From 69a6a321c2f70fd86dcbf6a196e4cf0196b03161 Mon Sep 17 00:00:00 2001
From: Magali Ruffier <mr6@ebi.ac.uk>
Date: Wed, 17 Jan 2018 13:06:47 +0000
Subject: [PATCH] make dbconnection persistent
---
.../xref_mapping/XrefMapper/BasicMapper.pm | 20 +-
.../xref_mapping/XrefMapper/ChecksumMapper.pm | 9 +-
.../XrefMapper/CoordinateMapper.pm | 1 +
.../xref_mapping/XrefMapper/DirectXrefs.pm | 108 ++++---
.../xref_mapping/XrefMapper/DisplayXrefs.pm | 102 ++++---
.../xref_mapping/XrefMapper/OfficialNaming.pm | 251 ++++++++-------
.../XrefMapper/ProcessMappings.pm | 52 ++--
.../xref_mapping/XrefMapper/ProcessPaired.pm | 45 ++-
.../XrefMapper/ProcessPrioritys.pm | 53 ++--
.../xref_mapping/XrefMapper/TestMappings.pm | 56 ++--
.../xref_mapping/XrefMapper/XrefLoader.pm | 146 ++++-----
.../XrefParser/ArrayExpressParser.pm | 12 +-
.../xref_mapping/XrefParser/BaseParser.pm | 289 ++++++++----------
.../xref_mapping/XrefParser/CCDSParser.pm | 5 +-
.../xref_mapping/XrefParser/CoreXrefParser.pm | 11 +-
.../xref_mapping/XrefParser/DBASSParser.pm | 9 +-
.../XrefParser/EntrezGeneParser.pm | 10 +-
.../xref_mapping/XrefParser/HGNCParser.pm | 36 ++-
.../XrefParser/HGNC_CCDSParser.pm | 9 +-
.../xref_mapping/XrefParser/HPAParser.pm | 7 +-
.../xref_mapping/XrefParser/JGI_Parser.pm | 6 +-
.../xref_mapping/XrefParser/MGIParser.pm | 10 +-
.../XrefParser/MGI_CCDS_Parser.pm | 5 +-
.../XrefParser/MGI_Desc_Parser.pm | 5 +-
.../xref_mapping/XrefParser/MIMParser.pm | 12 +-
.../xref_mapping/XrefParser/Mim2GeneParser.pm | 13 +-
.../xref_mapping/XrefParser/PomBaseParser.pm | 14 +-
.../xref_mapping/XrefParser/RFAMParser.pm | 14 +-
.../xref_mapping/XrefParser/RGDParser.pm | 13 +-
.../xref_mapping/XrefParser/ReactomeParser.pm | 24 +-
.../XrefParser/RefSeqCoordinateParser.pm | 27 +-
.../XrefParser/RefSeqGPFFParser.pm | 59 ++--
.../xref_mapping/XrefParser/RefSeqParser.pm | 36 +--
.../XrefParser/RefSeq_CCDSParser.pm | 11 +-
.../xref_mapping/XrefParser/SGDParser.pm | 14 +-
.../xref_mapping/XrefParser/UCSCParser.pm | 5 +-
.../XrefParser/UniProtDirectParser.pm | 8 +-
.../xref_mapping/XrefParser/UniProtParser.pm | 43 +--
.../UniProtParser_descriptions_only.pm | 30 +-
.../xref_mapping/XrefParser/VGNCParser.pm | 10 +-
.../XrefParser/XenopusJamboreeParser.pm | 3 +
.../xref_mapping/XrefParser/ZFINParser.pm | 15 +-
.../xref_mapping/XrefParser/miRBaseParser.pm | 13 +-
43 files changed, 862 insertions(+), 759 deletions(-)
diff --git a/misc-scripts/xref_mapping/XrefMapper/BasicMapper.pm b/misc-scripts/xref_mapping/XrefMapper/BasicMapper.pm
index 44f040e77a..7cfe346cd6 100644
--- a/misc-scripts/xref_mapping/XrefMapper/BasicMapper.pm
+++ b/misc-scripts/xref_mapping/XrefMapper/BasicMapper.pm
@@ -490,7 +490,6 @@ sub get_id_from_species_name {
while(my @row2 = $sth->fetchrow_array()){
print STDERR $row2[0]."\n";
}
- croak("Please try again :-)\n");
}
$sth->finish();
@@ -651,8 +650,8 @@ sub get_official_name {
# i.e. move all HGNC from transcripts to Genes.
#
sub biomart_fix{
- my ($self, $db_name, $type1, $type2, $verbose) = @_;
- my $xref_dbc = $self->xref->dbc;
+ my ($self, $db_name, $type1, $type2, $verbose, $xref_dbc) = @_;
+ $xref_dbc = $self->xref->dbc unless defined $xref_dbc;
print "$db_name is associated with both $type1 and $type2 object types\n" if(defined($verbose));
print "$db_name moved to Gene level.\n" if(!defined($verbose));
@@ -702,7 +701,6 @@ sub biomart_fix{
source.name = "$db_name";
EOF
my $result = $xref_dbc->do($sql) ;
-# print "\n$sql\n";
if($db_name eq "GO" || $db_name eq 'goslim_goa'){
$sql =(<<"EOF2");
@@ -867,7 +865,6 @@ sub get_species_id_from_species_name{
while(my @row2 = $sth->fetchrow_array()){
print STDERR $row2[0]."\n";
}
- croak("Please try again :-)\n");
}
$sth->finish();
@@ -1031,6 +1028,8 @@ sub get_alt_allele_hashes{
sub process_alt_alleles{
my $self = shift;
+ my $dbc = shift;
+ $dbc = $self->xref->dbc unless defined $dbc;
# ALL are on the Gene level now. This may change but for now it is okay.
my ($alt_to_ref, $ref_to_alts) = $self->get_alt_allele_hashes();
@@ -1086,9 +1085,9 @@ DELETE ox
DEL
$del_sql .= "'".join("', '",$self->get_gene_specific_list()) . "')";
- my $move_sth = $self->xref->dbc->prepare($move_sql) || croak "$move_sql cannot be prepared";
- my $del_ix_sth = $self->xref->dbc->prepare($del_ix_sql) || croak "$del_ix_sql cannot be prepared";
- my $del_sth = $self->xref->dbc->prepare($del_sql) || croak "$del_sql cannot be prepared";
+ my $move_sth = $dbc->prepare($move_sql) || croak "$move_sql cannot be prepared";
+ my $del_ix_sth = $dbc->prepare($del_ix_sql) || croak "$del_ix_sql cannot be prepared";
+ my $del_sth = $dbc->prepare($del_sql) || croak "$del_sql cannot be prepared";
my $move_count = 0;
my $del_ix_count = 0;
@@ -1229,14 +1228,15 @@ sub get_gene_specific_list {
#
sub source_defined_move{
my $self = shift;
+ my $dbi = shift;
my $tester = XrefMapper::TestMappings->new($self);
if($tester->unlinked_entries){
croak "Problems found before source_defined_move\n";
}
foreach my $source ($self->get_gene_specific_list()){
- $self->biomart_fix($source,"Translation","Gene");
- $self->biomart_fix($source,"Transcript","Gene");
+ $self->biomart_fix($source,"Translation","Gene", undef, undef, $dbi);
+ $self->biomart_fix($source,"Transcript","Gene", undef, undef, $dbi);
}
if($tester->unlinked_entries){
croak "Problems found after source_defined_move\n";
diff --git a/misc-scripts/xref_mapping/XrefMapper/ChecksumMapper.pm b/misc-scripts/xref_mapping/XrefMapper/ChecksumMapper.pm
index 26e7a8de8f..2f5d5707d9 100644
--- a/misc-scripts/xref_mapping/XrefMapper/ChecksumMapper.pm
+++ b/misc-scripts/xref_mapping/XrefMapper/ChecksumMapper.pm
@@ -71,7 +71,7 @@ sub target {
}
sub process {
- my ($self, $db_url) = @_;
+ my ($self, $db_url, $species_id) = @_;
$self->_update_status('checksum_xrefs_started');
my $source_id = $self->source_id();
@@ -82,7 +82,7 @@ sub process {
my $method = $self->get_method();
my $results = $method->run($target, $source_id, $object_type, $db_url);
$self->log_progress('Starting upload');
- $self->upload($results);
+ $self->upload($results, $species_id);
}
$self->_update_status('checksum_xrefs_finished');
@@ -90,7 +90,7 @@ sub process {
}
sub upload {
- my ($self, $results) = @_;
+ my ($self, $results, $species_id) = @_;
#The elements come in as an array looking like
# [ { id => 1, upi => 'UPI00000A', object_type => 'Translation' } ]
@@ -105,7 +105,8 @@ SQL
my $h = $self->_xref_helper();
my $source_id = $self->source_id();
- my $species_id = $self->species_id();
+ $species_id = $self->species_id() unless defined $species_id;
+ if (!defined $species_id) { return; }
$h->transaction(-CALLBACK => sub {
diff --git a/misc-scripts/xref_mapping/XrefMapper/CoordinateMapper.pm b/misc-scripts/xref_mapping/XrefMapper/CoordinateMapper.pm
index 9d95d622dd..bbc6d9d600 100644
--- a/misc-scripts/xref_mapping/XrefMapper/CoordinateMapper.pm
+++ b/misc-scripts/xref_mapping/XrefMapper/CoordinateMapper.pm
@@ -86,6 +86,7 @@ sub run_coordinatemapping {
my $species = $core_db->species();
# my $species_id = $self->mapper->core->species;
my $species_id = XrefMapper::BasicMapper::get_species_id_from_species_name( $xref_db, $species );
+ if (!defined $species_id) { return; }
# We only do coordinate mapping for mouse and human for now.
diff --git a/misc-scripts/xref_mapping/XrefMapper/DirectXrefs.pm b/misc-scripts/xref_mapping/XrefMapper/DirectXrefs.pm
index 76837d81c1..93c1477629 100644
--- a/misc-scripts/xref_mapping/XrefMapper/DirectXrefs.pm
+++ b/misc-scripts/xref_mapping/XrefMapper/DirectXrefs.pm
@@ -45,15 +45,14 @@ sub new {
sub get_ins_ix_sth {
my $self = shift;
+ my $dbi = shift;
- if(!defined($self->{'_ins_ix_sth'})){
- my $sql = (<<"IIX");
+ my $sql = (<<"IIX");
INSERT IGNORE INTO identity_xref (object_xref_id, query_identity, target_identity)
VALUES (?, 100, 100)
IIX
- $self->{'_ins_ix_sth'} = $self->xref->dbc->prepare($sql);
- }
- return $self->{'_ins_ix_sth'};
+ my $sth = $dbi->prepare($sql);
+ return $sth;
}
@@ -63,22 +62,19 @@ sub process {
# Now process the direct xrefs and add data to the object xrefs remember dependent xrefs.
my $object_xref_id;
- my $ox_sth = $self->xref->dbc->prepare("select max(object_xref_id) from object_xref");
- $ox_sth->execute();
- $ox_sth->bind_columns(\$object_xref_id);
- $ox_sth->fetch();
- $ox_sth->finish;
+ my $dbi = $self->xref->dbc;
# First get the sths needed for the processing of the direct xrefs;
my $ins_ox_sql = (<<"IOS");
-INSERT INTO object_xref (object_xref_id, ensembl_id, xref_id, ensembl_object_type, linkage_type)
- VALUES (?, ?, ?, ?, ?)
+INSERT INTO object_xref (ensembl_id, xref_id, ensembl_object_type, linkage_type)
+ VALUES (?, ?, ?, ?)
IOS
- my $ins_ox_sth = $self->xref->dbc->prepare($ins_ox_sql);
+ my $ins_ox_sth = $dbi->prepare($ins_ox_sql);
+ my $get_object_xref_id_sth = $self->get_ox_id_sth($dbi);
# Direct xrefs can be considered to be 100% matching
- my $ins_ix_sth = $self->get_ins_ix_sth();
+ my $ins_ix_sth = $self->get_ins_ix_sth($dbi);
my $stable_sql=(<<"SQL");
SELECT so.name, dx.general_xref_id, s.internal_id, dx.ensembl_stable_id , dx.linkage_xref
@@ -101,7 +97,7 @@ SQL
my ($dbname, $xref_id, $internal_id, $stable_id, $linkage_type);
my $sql = $stable_sql;
$sql =~ s/TYPE/$table/g;
- my $sth = $self->xref->dbc->prepare($sql);
+ my $sth = $dbi->prepare($sql);
$sth->execute();
$sth->bind_columns(\$dbname, \$xref_id, \$internal_id, \$stable_id, \$linkage_type);
my $count =0;
@@ -126,7 +122,9 @@ SQL
if($internal_id == 0){
die "Problem could not find stable id $stable_id and got past the first check for $dbname\n";
}
- $ins_ox_sth->execute($object_xref_id, $internal_id, $xref_id, $table, 'DIRECT');
+ $ins_ox_sth->execute($internal_id, $xref_id, $table, 'DIRECT');
+ $get_object_xref_id_sth->execute($internal_id, $xref_id, $table, 'DIRECT');
+ $object_xref_id = ($get_object_xref_id_sth->fetchrow_array())[0];
if($ins_ox_sth->err){
$duplicate_direct_count++;
next; #duplicate
@@ -136,10 +134,10 @@ SQL
push @master_xref_ids, $xref_id;
}
$self->process_dependents({master_xrefs => \@master_xref_ids,
- max_object_xref_id => \$object_xref_id,
dup_count => \$duplicate_dependent_count,
table => $table,
- internal_id => $internal_id
+ internal_id => $internal_id,
+ dbi => $dbi,
});
}
@@ -152,7 +150,7 @@ SQL
print STDERR "*WARNING*: ".$err_count{$key}." direct xrefs for database ".$key." could not be added as their stable_ids could not be found\n";
}
- my $sth = $self->xref->dbc->prepare("insert into process_status (status, date) values('direct_xrefs_parsed',now())");
+ my $sth = $dbi->prepare("insert into process_status (status, date) values('direct_xrefs_parsed',now())");
$sth->execute();
$sth->finish;
@@ -162,72 +160,94 @@ SQL
sub get_dep_sth {
my $self = shift;
+ my $dbi = shift;
- if(!defined($self->{'_dep_sth'})){
- my $dep_sql = (<<"DSS");
+ my $dep_sql = (<<"DSS");
SELECT dependent_xref_id, linkage_annotation
FROM dependent_xref
WHERE master_xref_id = ?
DSS
- $self->{'_dep_sth'} = $self->xref->dbc->prepare($dep_sql);
- }
- return $self->{'_dep_sth'};
+ my $sth = $dbi->prepare($dep_sql);
+ return $sth;
}
sub get_dep_go_sth {
my $self = shift;
+ my $dbi = shift;
- if(!defined($self->{'_dep_go_sth'})){
- my $sql = (<<"IGO");
+ my $sql = (<<"IGO");
INSERT INTO go_xref (object_xref_id, linkage_type, source_xref_id)
VALUES (?,?,?)
IGO
- $self->{'_dep_go_sth'} = $self->xref->dbc->prepare($sql);
- }
- return $self->{'_dep_go_sth'};
+ my $sth = $dbi->prepare($sql);
+ return $sth;
}
sub get_add_dep_ox {
my $self = shift;
+ my $dbi = shift;
- if(!defined($self->{'_add_dep_ox_sth'})){
- my $sql = (<<"IO2");
-INSERT INTO object_xref (object_xref_id, ensembl_id, xref_id, ensembl_object_type, linkage_type, master_xref_id)
- VALUES (?, ?, ?, ?, ?, ?)
+ my $sql = (<<"IO2");
+INSERT INTO object_xref (ensembl_id, xref_id, ensembl_object_type, linkage_type, master_xref_id)
+ VALUES (?, ?, ?, ?, ?)
IO2
- $self->{'_add_dep_ox_sth'} = $self->xref->dbc->prepare($sql);
- }
- return $self->{'_add_dep_ox_sth'};
+ my $sth = $dbi->prepare($sql);
+ return $sth;
+}
+
+sub get_ox_id_sth {
+ my $self = shift;
+ my $dbi = shift;
+
+ my $sql = (<<"IO2");
+select object_xref_id from object_xref where ensembl_id = ? and xref_id = ? and ensembl_object_type = ? and linkage_type = ?
+IO2
+ my $sth = $dbi->prepare($sql);
+ return $sth;
+}
+
+sub get_ox_id_master_sth {
+ my $self = shift;
+ my $dbi = shift;
+
+ my $sql = (<<"IO2");
+select object_xref_id from object_xref where ensembl_id = ? and xref_id = ? and ensembl_object_type = ? and linkage_type = ? and master_xref_id = ?
+IO2
+ my $sth = $dbi->prepare($sql);
+ return $sth;
}
sub process_dependents {
my ($self, $arg_ref) = @_;
+ my $dbi = $arg_ref->{dbi};
my $master_xref_ids = $arg_ref->{master_xrefs};
- my $object_xref_id = $arg_ref->{max_object_xref_id};
my $duplicate_dep_count = $arg_ref->{dup_count};
my $table = $arg_ref->{table};
my $internal_id = $arg_ref->{internal_id};
- my $dep_sth = $self->get_dep_sth();
- my $ins_go_dep_sth = $self->get_dep_go_sth();
- my $ins_ox_sth2 = $self->get_add_dep_ox();
- my $ins_ix_sth = $self->get_ins_ix_sth();
+ my $dep_sth = $self->get_dep_sth($dbi);
+ my $ins_go_dep_sth = $self->get_dep_go_sth($dbi);
+ my $ins_ox_sth2 = $self->get_add_dep_ox($dbi);
+ my $ins_ix_sth = $self->get_ins_ix_sth($dbi);
+ my $get_object_xref_id_sth = $self->get_ox_id_master_sth($dbi);
local $ins_ox_sth2->{RaiseError} = 0; # want to see duplicates and not die automatically
local $ins_ox_sth2->{PrintError} = 0;
+ my $object_xref_id;
while(my $master_xref_id = pop(@$master_xref_ids)){
my ($dep_xref_id, $link);
$dep_sth->execute($master_xref_id);
$dep_sth->bind_columns(\$dep_xref_id, \$link);
while($dep_sth->fetch){
- $$object_xref_id++;
- $ins_ox_sth2->execute($$object_xref_id, $internal_id, $dep_xref_id, $table, 'DEPENDENT', $master_xref_id);
+ $ins_ox_sth2->execute($internal_id, $dep_xref_id, $table, 'DEPENDENT', $master_xref_id);
+ $get_object_xref_id_sth->execute($internal_id, $dep_xref_id, $table, 'DEPENDENT', $master_xref_id);
+ $object_xref_id = ($get_object_xref_id_sth->fetchrow_array())[0];
if($ins_ox_sth2->err){
my $err = $ins_ox_sth2->errstr;
if($err =~ /Duplicate/){
@@ -238,7 +258,7 @@ sub process_dependents {
die "Problem loading error is $err\n";
}
}
- $ins_ix_sth->execute($$object_xref_id);
+ $ins_ix_sth->execute($object_xref_id);
push @$master_xref_ids, $dep_xref_id; # get the dependent, dependents just in case
if(defined($link) and $link ne ""){ # we have a go term linkage type
diff --git a/misc-scripts/xref_mapping/XrefMapper/DisplayXrefs.pm b/misc-scripts/xref_mapping/XrefMapper/DisplayXrefs.pm
index 561280c30b..0b0b8d41dd 100644
--- a/misc-scripts/xref_mapping/XrefMapper/DisplayXrefs.pm
+++ b/misc-scripts/xref_mapping/XrefMapper/DisplayXrefs.pm
@@ -213,25 +213,27 @@ sub set_display_xrefs_from_stable_table{
print "Setting Transcript and Gene display_xrefs from xref database into core and setting the desc\n" if ($self->verbose);
my $xref_offset = $self->get_meta_value("xref_offset");
+ my $core_dbi = $self->core->dbc;
+ my $xref_dbi = $self->xref->dbc;
print "Using xref_off set of $xref_offset\n" if($self->verbose);
- my $reset_sth = $self->core->dbc->prepare("UPDATE gene SET display_xref_id = null");
+ my $reset_sth = $core_dbi->prepare("UPDATE gene SET display_xref_id = null");
$reset_sth->execute();
$reset_sth->finish;
- $reset_sth = $self->core->dbc->prepare("UPDATE transcript SET display_xref_id = null WHERE biotype NOT IN ('LRG_gene')");
+ $reset_sth = $core_dbi->prepare("UPDATE transcript SET display_xref_id = null WHERE biotype NOT IN ('LRG_gene')");
$reset_sth->execute();
$reset_sth->finish;
- $reset_sth = $self->core->dbc->prepare("UPDATE gene SET description = null");
+ $reset_sth = $core_dbi->prepare("UPDATE gene SET description = null");
$reset_sth->execute();
$reset_sth->finish;
my %name_to_external_name;
my $sql = "select external_db_id, db_name, db_display_name from external_db";
- my $sth = $self->core->dbc->prepare($sql);
+ my $sth = $core_dbi->prepare($sql);
$sth->execute();
my ($id, $name, $display_name);
$sth->bind_columns(\$id, \$name, \$display_name);
@@ -243,7 +245,7 @@ sub set_display_xrefs_from_stable_table{
my %source_id_to_external_name;
$sql = 'select s.source_id, s.name from source s, xref x where x.source_id = s.source_id group by s.source_id'; # only get those of interest
- $sth = $self->xref->dbc->prepare($sql);
+ $sth = $xref_dbi->prepare($sql);
$sth->execute();
$sth->bind_columns(\$id, \$name);
@@ -255,21 +257,21 @@ sub set_display_xrefs_from_stable_table{
$sth->finish;
- my $update_gene_sth = $self->core->dbc->prepare("UPDATE gene g SET g.display_xref_id= ? WHERE g.gene_id=?");
- my $update_gene_desc_sth = $self->core->dbc->prepare("UPDATE gene g SET g.description= ? WHERE g.gene_id=?");
+ my $update_gene_sth = $core_dbi->prepare("UPDATE gene g SET g.display_xref_id= ? WHERE g.gene_id=?");
+ my $update_gene_desc_sth = $core_dbi->prepare("UPDATE gene g SET g.description= ? WHERE g.gene_id=?");
- my $update_tran_sth = $self->core->dbc->prepare("UPDATE transcript t SET t.display_xref_id= ? WHERE t.transcript_id=?");
+ my $update_tran_sth = $core_dbi->prepare("UPDATE transcript t SET t.display_xref_id= ? WHERE t.transcript_id=?");
- my $get_gene_display_xref = $self->xref->dbc->prepare("SELECT gsi.internal_id, gsi.display_xref_id, x.description ,x.source_id, x.accession
+ my $get_gene_display_xref = $xref_dbi->prepare("SELECT gsi.internal_id, gsi.display_xref_id, x.description ,x.source_id, x.accession
FROM gene_stable_id gsi, xref x
WHERE gsi.display_xref_id = x.xref_id");
- my $get_tran_display_xref = $self->xref->dbc->prepare("SELECT gsi.internal_id, gsi.display_xref_id from transcript_stable_id gsi");
+ my $get_tran_display_xref = $xref_dbi->prepare("SELECT gsi.internal_id, gsi.display_xref_id from transcript_stable_id gsi");
- $reset_sth = $self->xref->dbc->prepare("UPDATE gene_stable_id gsi SET gsi.desc_set=0");
+ $reset_sth = $xref_dbi->prepare("UPDATE gene_stable_id gsi SET gsi.desc_set=0");
$reset_sth->execute();
- my $set_desc_done_sth = $self->xref->dbc->prepare("UPDATE gene_stable_id gsi SET gsi.desc_set=1 WHERE gsi.internal_id=?");
+ my $set_desc_done_sth = $xref_dbi->prepare("UPDATE gene_stable_id gsi SET gsi.desc_set=1 WHERE gsi.internal_id=?");
$get_gene_display_xref->execute();
my $xref_id;
@@ -315,7 +317,7 @@ sub set_display_xrefs_from_stable_table{
# Synonyms are only used as alternative gene names, so should be synonyms of the gene symbol chosen
#
- my $syn_clean_sth = $self->core->dbc->prepare("DELETE es FROM external_synonym es, xref x LEFT JOIN gene g ON g.display_xref_id = x.xref_id WHERE es.xref_id = x.xref_id AND isnull(g.display_xref_id)");
+ my $syn_clean_sth = $core_dbi->prepare("DELETE es FROM external_synonym es, xref x LEFT JOIN gene g ON g.display_xref_id = x.xref_id WHERE es.xref_id = x.xref_id AND isnull(g.display_xref_id)");
$syn_clean_sth->execute();
$syn_clean_sth->finish();
@@ -384,35 +386,37 @@ sub set_display_xrefs{
print "Building Transcript and Gene display_xrefs\n" if ($self->verbose);
my $xref_offset = $self->get_meta_value("xref_offset");
+ my $core_dbi = $self->core->dbc();
+ my $xref_dbi = $self->xref->dbc();
print "Using xref_off set of $xref_offset\n" if($self->verbose);
- my $reset_sth = $self->core->dbc->prepare("UPDATE gene SET display_xref_id = null");
+ my $reset_sth = $core_dbi->prepare("UPDATE gene SET display_xref_id = null");
$reset_sth->execute();
$reset_sth->finish;
- $reset_sth = $self->core->dbc->prepare("UPDATE transcript SET display_xref_id = null WHERE biotype NOT IN ('LRG_gene')");
+ $reset_sth = $core_dbi->prepare("UPDATE transcript SET display_xref_id = null WHERE biotype NOT IN ('LRG_gene')");
$reset_sth->execute();
$reset_sth->finish;
- my $update_gene_sth = $self->core->dbc->prepare("UPDATE gene g SET g.display_xref_id= ? WHERE g.gene_id=?");
- my $update_tran_sth = $self->core->dbc->prepare("UPDATE transcript t SET t.display_xref_id= ? WHERE t.transcript_id=?");
+ my $update_gene_sth = $core_dbi->prepare("UPDATE gene g SET g.display_xref_id= ? WHERE g.gene_id=?");
+ my $update_tran_sth = $core_dbi->prepare("UPDATE transcript t SET t.display_xref_id= ? WHERE t.transcript_id=?");
# Set status to 'NO_DISPLAY' for object_xrefs with a display_label that is just numeric;
- my $update_ignore_sth = $self->xref->dbc->prepare('UPDATE object_xref ox, source s, xref x SET ox_status = "NO_DISPLAY" where ox_status like "DUMP_OUT" and s.source_id = x.source_id and x.label REGEXP "^[0-9]+$" and ox.xref_id = x.xref_id');
+ my $update_ignore_sth = $xref_dbi->prepare('UPDATE object_xref ox, source s, xref x SET ox_status = "NO_DISPLAY" where ox_status like "DUMP_OUT" and s.source_id = x.source_id and x.label REGEXP "^[0-9]+$" and ox.xref_id = x.xref_id');
$update_ignore_sth->execute();
$update_ignore_sth->finish;
- my $ins_p_sth = $self->xref->dbc->prepare("INSERT ignore into display_xref_priority (ensembl_object_type,source_id, priority) values(?, ?, ?)");
- my $get_source_id_sth = $self->xref->dbc->prepare("select source_id from source where name like ? order by priority");
- my $list_sources_sth = $self->xref->dbc->prepare("select distinct name from display_xref_priority d join source using(source_id) where ensembl_object_type = ? order by d.priority");
+ my $ins_p_sth = $xref_dbi->prepare("INSERT ignore into display_xref_priority (ensembl_object_type,source_id, priority) values(?, ?, ?)");
+ my $get_source_id_sth = $xref_dbi->prepare("select source_id from source where name like ? order by priority");
+ my $list_sources_sth = $xref_dbi->prepare("select distinct name from display_xref_priority d join source using(source_id) where ensembl_object_type = ? order by d.priority");
- $update_ignore_sth = $self->xref->dbc->prepare('UPDATE object_xref SET ox_status = "NO_DISPLAY" where object_xref_id = ?');
+ $update_ignore_sth = $xref_dbi->prepare('UPDATE object_xref SET ox_status = "NO_DISPLAY" where object_xref_id = ?');
my %object_types = ('gene' => 'Gene', 'transcript' => 'Transcript');
@@ -461,7 +465,7 @@ sub set_display_xrefs{
foreach my $ignore_sql (values %$ignore){
print "IGNORE SQL: $ignore_sql\n" if($self->verbose);
- my $ignore_sth = $self->xref->dbc->prepare($ignore_sql);
+ my $ignore_sth = $xref_dbi->prepare($ignore_sql);
$ignore_sth->execute();
my ($object_xref_id);
$ignore_sth->bind_columns(\$object_xref_id);
@@ -518,7 +522,7 @@ DXS
my %object_seen;
- my $display_xref_sth = $self->xref->dbc->prepare($display_xref_sql);
+ my $display_xref_sth = $xref_dbi->prepare($display_xref_sql);
my $display_xref_count = 0;
$display_xref_sth->execute($object_type);
@@ -555,7 +559,7 @@ DXS
# reset the status to DUMP_OUT fro object_xrefs that where ignored for the display_xref;
#
- my $reset_status_sth = $self->xref->dbc->prepare('UPDATE object_xref SET ox_status = "DUMP_OUT" where ox_status = "NO_DISPLAY"');
+ my $reset_status_sth = $xref_dbi->prepare('UPDATE object_xref SET ox_status = "DUMP_OUT" where ox_status = "NO_DISPLAY"');
$reset_status_sth->execute();
$reset_status_sth->finish;
@@ -564,7 +568,7 @@ DXS
# Synonyms are only used as alternative gene names, so should be synonyms of the gene symbol chosen
#
- my $syn_clean_sth = $self->core->dbc->prepare("DELETE es FROM external_synonym es, xref x LEFT JOIN gene g ON g.display_xref_id = x.xref_id WHERE es.xref_id = x.xref_id AND isnull(g.display_xref_id)");
+ my $syn_clean_sth = $core_dbi->prepare("DELETE es FROM external_synonym es, xref x LEFT JOIN gene g ON g.display_xref_id = x.xref_id WHERE es.xref_id = x.xref_id AND isnull(g.display_xref_id)");
$syn_clean_sth->execute();
$syn_clean_sth->finish();
@@ -574,25 +578,27 @@ DXS
sub transcript_names_from_gene {
my $self = shift;
+ my $core_dbi = $self->core->dbc;
+ my $xref_dbi = $self->xref->dbc;
print "Assigning transcript names from gene names\n" if ($self->verbose);
- my $reset_sth = $self->core->dbc->prepare("UPDATE transcript SET display_xref_id = null WHERE biotype NOT IN ('LRG_gene')");
+ my $reset_sth = $core_dbi->prepare("UPDATE transcript SET display_xref_id = null WHERE biotype NOT IN ('LRG_gene')");
$reset_sth->execute();
$reset_sth->finish;
- my $xref_id_sth = $self->core->dbc->prepare("SELECT max(xref_id) FROM xref");
- my $ox_id_sth = $self->core->dbc->prepare("SELECT max(object_xref_id) FROM object_xref");
- my $del_xref_sth = $self->core->dbc->prepare("DELETE x FROM xref x, object_xref ox WHERE x.xref_id = ox.xref_id AND ensembl_object_type = 'Transcript' AND display_label REGEXP '-2[0-9]{2}\$'");
- my $reuse_xref_sth = $self->core->dbc->prepare("SELECT xref_id FROM xref x WHERE external_db_id = ? AND display_label = ? AND description = ? AND info_type = 'MISC'");
- my $del_ox_sth = $self->core->dbc->prepare("DELETE ox FROM object_xref ox LEFT JOIN xref x ON x.xref_id = ox.xref_id WHERE isnull(x.xref_id)");
- my $ins_xref_sth = $self->core->dbc->prepare("INSERT IGNORE into xref (xref_id, external_db_id, dbprimary_acc, display_label, version, description, info_type, info_text) values(?, ?, ?, ?, 0, ?, 'MISC', ?)");
- my $ins_ox_sth = $self->core->dbc->prepare("INSERT into object_xref (object_xref_id, ensembl_id, ensembl_object_type, xref_id) values(?, ?, 'Transcript', ?)");
- my $update_tran_sth = $self->core->dbc->prepare("UPDATE transcript t SET t.display_xref_id= ? WHERE t.transcript_id=?");
+ my $xref_id_sth = $core_dbi->prepare("SELECT max(xref_id) FROM xref");
+ my $ox_id_sth = $core_dbi->prepare("SELECT max(object_xref_id) FROM object_xref");
+ my $del_xref_sth = $core_dbi->prepare("DELETE x FROM xref x, object_xref ox WHERE x.xref_id = ox.xref_id AND ensembl_object_type = 'Transcript' AND display_label REGEXP '-2[0-9]{2}\$'");
+ my $reuse_xref_sth = $core_dbi->prepare("SELECT xref_id FROM xref x WHERE external_db_id = ? AND display_label = ? AND description = ? AND info_type = 'MISC'");
+ my $del_ox_sth = $core_dbi->prepare("DELETE ox FROM object_xref ox LEFT JOIN xref x ON x.xref_id = ox.xref_id WHERE isnull(x.xref_id)");
+ my $ins_xref_sth = $core_dbi->prepare("INSERT IGNORE into xref (xref_id, external_db_id, dbprimary_acc, display_label, version, description, info_type, info_text) values(?, ?, ?, ?, 0, ?, 'MISC', ?)");
+ my $ins_ox_sth = $core_dbi->prepare("INSERT into object_xref (object_xref_id, ensembl_id, ensembl_object_type, xref_id) values(?, ?, 'Transcript', ?)");
+ my $update_tran_sth = $core_dbi->prepare("UPDATE transcript t SET t.display_xref_id= ? WHERE t.transcript_id=?");
- my $get_genes = $self->core->dbc->prepare("SELECT g.gene_id, e.db_name, x.dbprimary_acc, x.display_label, x.description FROM gene g, xref x, external_db e where g.display_xref_id = x.xref_id and e.external_db_id = x.external_db_id");
- my $get_transcripts = $self->core->dbc->prepare("SELECT transcript_id FROM transcript WHERE gene_id = ? ORDER BY seq_region_start, seq_region_end");
- my $get_source_id = $self->core->dbc->prepare("SELECT external_db_id FROM external_db WHERE db_name like ?");
+ my $get_genes = $core_dbi->prepare("SELECT g.gene_id, e.db_name, x.dbprimary_acc, x.display_label, x.description FROM gene g, xref x, external_db e where g.display_xref_id = x.xref_id and e.external_db_id = x.external_db_id");
+ my $get_transcripts = $core_dbi->prepare("SELECT transcript_id FROM transcript WHERE gene_id = ? ORDER BY seq_region_start, seq_region_end");
+ my $get_source_id = $core_dbi->prepare("SELECT external_db_id FROM external_db WHERE db_name like ?");
$get_genes->execute();
my ($gene_id, $external_db, $external_db_id, $acc, $label, $description, $transcript_id, $xref_id, $ox_id, $ext, $reuse_xref_id, $info_text);
@@ -701,11 +707,13 @@ sub set_gene_descriptions{
my $self = shift;
my $only_those_not_set = shift || 0;
my $sql;
+ my $core_dbi = $self->core->dbc;
+ my $xref_dbi = $self->xref->dbc;
- my $update_gene_desc_sth = $self->core->dbc->prepare("UPDATE gene SET description = ? where gene_id = ?");
+ my $update_gene_desc_sth = $core_dbi->prepare("UPDATE gene SET description = ? where gene_id = ?");
if(!$only_those_not_set){
- my $reset_sth = $self->core->dbc->prepare("UPDATE gene SET description = null");
+ my $reset_sth = $core_dbi->prepare("UPDATE gene SET description = null");
$reset_sth->execute();
$reset_sth->finish;
}
@@ -714,7 +722,7 @@ sub set_gene_descriptions{
if($only_those_not_set){
print "Only setting those not already set\n";
$sql = "select internal_id from gene_stable_id where desc_set = 1";
- my $sql_sth = $self->xref->dbc->prepare($sql);
+ my $sql_sth = $xref_dbi->prepare($sql);
$sql_sth->execute;
my $id;
$sql_sth->bind_columns(\$id);
@@ -730,7 +738,7 @@ sub set_gene_descriptions{
my %name_to_external_name;
$sql = "select external_db_id, db_name, db_display_name from external_db";
- my $sth = $self->core->dbc->prepare($sql);
+ my $sth = $core_dbi->prepare($sql);
$sth->execute();
my ($id, $name, $display_name);
$sth->bind_columns(\$id, \$name, \$display_name);
@@ -767,9 +775,9 @@ sub set_gene_descriptions{
@regexps = $self->gene_description_filter_regexps();
}
- my $ins_p_sth = $self->xref->dbc->prepare("INSERT into gene_desc_priority (source_id, priority) values(?, ?)");
- my $get_source_id_sth = $self->xref->dbc->prepare("select source_id from source where name like ?");
- my $list_sources_sth = $self->xref->dbc->prepare("select distinct name from gene_desc_priority d join source using(source_id) order by d.priority");
+ my $ins_p_sth = $xref_dbi->prepare("INSERT ignore into gene_desc_priority (source_id, priority) values(?, ?)");
+ my $get_source_id_sth = $xref_dbi->prepare("select source_id from source where name like ?");
+ my $list_sources_sth = $xref_dbi->prepare("select distinct name from gene_desc_priority d join source using(source_id) order by d.priority");
# The lower the priority number the better then
my $i=0;
@@ -837,7 +845,7 @@ DXS
########################################################################
- my $gene_sth = $self->core->dbc->prepare("select g.description from gene g where g.gene_id = ?");
+ my $gene_sth = $core_dbi->prepare("select g.description from gene g where g.gene_id = ?");
my %no_source_name_in_desc;
if( $self->mapper->can("no_source_label_list") ){
@@ -848,7 +856,7 @@ DXS
}
}
- my $gene_desc_sth = $self->xref->dbc->prepare($gene_desc_sql);
+ my $gene_desc_sth = $xref_dbi->prepare($gene_desc_sql);
$gene_desc_sth->execute();
my ($gene_id, $desc,$source_id,$label);
diff --git a/misc-scripts/xref_mapping/XrefMapper/OfficialNaming.pm b/misc-scripts/xref_mapping/XrefMapper/OfficialNaming.pm
index cfb5e42b67..65227c7439 100644
--- a/misc-scripts/xref_mapping/XrefMapper/OfficialNaming.pm
+++ b/misc-scripts/xref_mapping/XrefMapper/OfficialNaming.pm
@@ -114,8 +114,10 @@ sub get_official_name {
##################################################
sub run {
my $self = shift;
+ my $species_id = shift;
my $dbname = $self->get_official_name();
+ my $dbi = $self->xref->dbc;
###########################################################
# If no offical name then we do not want to go any further
@@ -125,7 +127,8 @@ sub run {
$self->update_process_status("official_naming_done");
return;
}
- $self->species_id($self->get_id_from_species_name($self->core->species));
+ $species_id = $self->get_id_from_species_name($self->core->species) unless defined $species_id;
+ $self->species_id($species_id);
###########################################################
@@ -157,30 +160,30 @@ sub run {
######################################################
# Get the current max values for xref and object_xref
######################################################
- my ($max_object_xref_id, $max_xref_id) = $self->find_max_ids();
+ my ($max_object_xref_id, $max_xref_id) = $self->find_max_ids($dbi);
my %display_label_to_desc;
- $self->get_display_label_data(\%display_label_to_desc);
+ $self->get_display_label_data(\%display_label_to_desc, $dbi);
my %synonym;
- $self->get_synonyms(\%synonym);
+ $self->get_synonyms(\%synonym, $dbi);
# get the officail naming external_sources
- my $dbname_to_source_id = $self->get_new_dbname_sources(); # reference to hash
+ my $dbname_to_source_id = $self->get_new_dbname_sources($dbi); # reference to hash
###########################
# Delete the old ones.
###########################
- $self->delete_old_data($dbname_to_source_id);
+ $self->delete_old_data($dbname_to_source_id, $dbi);
#
# Different to to the other names above as not all will be new ones.
# For vega tran entries we only want to delete the new ones, those having 'MISC' as info_type;
#
- $self->clean_up_vega_tran_entries($dbname_to_source_id->{"Clone_based_vega_transcript"});
+ $self->clean_up_vega_tran_entries($dbname_to_source_id->{"Clone_based_vega_transcript"}, $dbi);
- $self->reset_display_xrefs();
+ $self->reset_display_xrefs($dbi);
my $db = new Bio::EnsEMBL::DBSQL::DBAdaptor(-dbconn => $self->core->dbc);
my $ga = $db->get_GeneAdaptor();
@@ -197,7 +200,7 @@ SELECT gtt.gene_id, gtt.transcript_id, gsi.stable_id, tsi.stable_id
ORDER BY gsi.stable_id, tsi.stable_id
SQ0
- my $sth = $self->xref->dbc->prepare($sql);
+ my $sth = $dbi->prepare($sql);
$sth->execute;
my ($gene_id, $tran_id, $gsi, $tsi);
@@ -212,11 +215,11 @@ SQ0
$tran_id_to_stable_id{$tran_id} = $tsi;
}
- my $dbentrie_sth = $self->get_dbentrie_sth();
- my $ins_xref_sth = $self->get_ins_xref_sth();
- my $ins_dep_ix_sth = $self->get_ins_dep_ix_sth();
- my $ins_object_xref_sth = $self->get_ins_object_xref_sth();
- my $set_gene_display_xref_sth = $self->get_set_gene_display_xref_sth();
+ my $dbentrie_sth = $self->get_dbentrie_sth($dbi);
+ my $ins_xref_sth = $self->get_ins_xref_sth($dbi);
+ my $ins_dep_ix_sth = $self->get_ins_dep_ix_sth($dbi);
+ my $ins_object_xref_sth = $self->get_ins_object_xref_sth($dbi);
+ my $set_gene_display_xref_sth = $self->get_set_gene_display_xref_sth($dbi);
my %xref_added; # store those added $xref_added{$accession:$source_id} = $xref_id;
my %seen_gene;
@@ -240,7 +243,7 @@ SQ0
IEG
my %ignore_object;
- my $ignore_sth = $self->xref->dbc->prepare($ignore_sql);
+ my $ignore_sth = $dbi->prepare($ignore_sql);
$ignore_sth->execute();
my ($ignore_object_xref_id);
$ignore_sth->bind_columns(\$ignore_object_xref_id);
@@ -269,7 +272,8 @@ IEG
label_to_desc => \%display_label_to_desc,
synomnym => \%synonym,
xref_added => \%xref_added,
- cbvt => $dbname_to_source_id->{"Clone_based_vega_transcript"}
+ cbvt => $dbname_to_source_id->{"Clone_based_vega_transcript"},
+ dbi => $dbi,
});
if (!defined($ens_clone_genes{$gene_id})) { #we're processing this gene for the first time
@@ -282,7 +286,8 @@ IEG
gene_to_tran => \%gene_to_transcripts,
tran_to_vega_name => $tran_to_vega_name,
gene_id_to_stable_id => \%gene_id_to_stable_id,
- official_name_used => \%official_name_used
+ official_name_used => \%official_name_used,
+ dbi => $dbi
});
if (defined($gene_symbol_xref_id)) {
@@ -293,7 +298,7 @@ IEG
# If not found see if there is an LRG entry
############################################
if(!defined($gene_symbol)){ # look for LRG
- ($gene_symbol, $gene_symbol_xref_id, $is_lrg) = $self->find_lrg_hgnc($gene_id);
+ ($gene_symbol, $gene_symbol_xref_id, $is_lrg) = $self->find_lrg_hgnc($gene_id, $dbi);
}
####################################################
@@ -305,6 +310,7 @@ IEG
$self->find_from_other_sources(\%ignore_object,
{gene_id => $gene_id,
label_to_desc => \%display_label_to_desc,
+ dbi => $dbi,
tran_source => \$tran_source});
}
@@ -350,7 +356,7 @@ IEG
my $desc = $display_label_to_desc{$gene_symbol};
if(!defined($gene_symbol_xref_id)){
- $gene_symbol_xref_id = $self->get_xref_id_for_gene_symbol($gene_symbol, $dbname);
+ $gene_symbol_xref_id = $self->get_xref_id_for_gene_symbol($gene_symbol, $dbname, $dbi);
if(!defined($gene_symbol_xref_id)){
carp "BLOOMING NORA could not find $gene_symbol in $dbname\n";
next;
@@ -365,6 +371,7 @@ IEG
gene_id_to_stable_id => \%gene_id_to_stable_id,
gene_symbol => $gene_symbol,
desc => $desc,
+ dbi => $dbi,
source_id => $dbname_to_source_id->{$tran_source."_trans_name"},
xref_added => \%xref_added,
seen_gene => \%seen_gene,
@@ -385,6 +392,7 @@ IEG
max_xref => \$max_xref_id,
max_object => \$max_object_xref_id,
xref_added => \%xref_added,
+ dbi => $dbi,
gene_to_tran => \%gene_to_transcripts,
tran_to_vega_ext => $tran_to_vega_ext,
ens_clone_genes => \%ens_clone_genes,
@@ -423,6 +431,7 @@ sub get_official_domain_name{
my $tran_to_vega_name = $arg_ref->{tran_to_vega_name};
my $gene_to_transcripts = $arg_ref->{gene_to_tran};
my $official_name_used = $arg_ref->{official_name_used};
+ my $dbi = $arg_ref->{dbi};
my $dbname = $self->get_official_name();
@@ -430,7 +439,7 @@ sub get_official_domain_name{
my $gene_symbol_xref_id = undef;
- my $dbentrie_sth = $self->get_dbentrie_sth();
+ my $dbentrie_sth = $self->get_dbentrie_sth($dbi);
my %ODN=();
my %xref_id_to_display;
@@ -462,7 +471,7 @@ sub get_official_domain_name{
if(($count > 1) and (scalar(keys %ODN) == 1)){ # found one that is "best" so set it and remove others
print "For gene ".$gene_id_to_stable_id->{$gene_id}." we have mutiple ".$dbname."'s\n";
- ($gene_symbol, $gene_symbol_xref_id) = $self->set_the_best_odns(\%ODN, \@list, \@list_ox, \%xref_id_to_display);
+ ($gene_symbol, $gene_symbol_xref_id) = $self->set_the_best_odns(\%ODN, \@list, \@list_ox, \%xref_id_to_display, $dbi);
if(defined($gene_symbol)){
return $gene_symbol, $gene_symbol_xref_id;
}
@@ -474,7 +483,7 @@ sub get_official_domain_name{
if(scalar(keys %ODN) > 1){
#if we have more than 1 xref, fail xrefs with worse % identity if we can (query or target identity whichever is greater)
- my $identity_sth = $self->get_best_identity_sth();
+ my $identity_sth = $self->get_best_identity_sth($dbi);
$identity_sth->execute($dbname, $gene_id, "Gene");
my ($xref_id, $best_identity);
$identity_sth->bind_columns(\$xref_id, \$best_identity);
@@ -537,7 +546,7 @@ sub get_official_domain_name{
%ODN = %best_ids;
print "For gene ".$gene_id_to_stable_id->{$gene_id}." we have mutiple ".$dbname."'s\n";
#set statuses for xrefs with worse % identity to MULTI_DELETE
- ($gene_symbol, $gene_symbol_xref_id) = $self->set_the_best_odns(\%ODN, \@list, \@list_ox, \%xref_id_to_display);
+ ($gene_symbol, $gene_symbol_xref_id) = $self->set_the_best_odns(\%ODN, \@list, \@list_ox, \%xref_id_to_display, $dbi);
if( defined($gene_symbol) && scalar(keys %ODN == 1) ){
return $gene_symbol, $gene_symbol_xref_id;
}
@@ -602,12 +611,13 @@ sub set_transcript_and_gene_display_xref_via_clone_name{
my $vega_clone_name = $arg_ref->{vega_clone_name};
my $clone_name = $arg_ref->{clone_name};
my $ens_clone_names = $arg_ref->{ens_clone_genes};
+ my $dbi = $arg_ref->{dbi};
- my $ins_xref_sth = $self->get_ins_xref_sth();
- my $ins_dep_ix_sth = $self->get_ins_dep_ix_sth();
- my $set_tran_display_xref_sth = $self->get_set_transcript_display_xref_sth();
- my $ins_object_xref_sth = $self->get_ins_object_xref_sth();
- my $set_gene_display_xref_sth = $self->get_set_gene_display_xref_sth();
+ my $ins_xref_sth = $self->get_ins_xref_sth($dbi);
+ my $ins_dep_ix_sth = $self->get_ins_dep_ix_sth($dbi);
+ my $set_tran_display_xref_sth = $self->get_set_transcript_display_xref_sth($dbi);
+ my $ins_object_xref_sth = $self->get_ins_object_xref_sth($dbi);
+ my $set_gene_display_xref_sth = $self->get_set_gene_display_xref_sth($dbi);
my $keep_gene;
@@ -719,13 +729,14 @@ sub set_transcript_display_xrefs{
my $tran_to_vega_ext = $arg_ref->{tran_to_vega_ext};
my $tran_source = $arg_ref->{tran_source};
my $gene_id_to_stable_id = $arg_ref->{gene_id_to_stable_id};
+ my $dbi = $arg_ref->{dbi};
# statement handles needed
- my $ins_xref_sth = $self->get_ins_xref_sth();
- my $ins_dep_ix_sth = $self->get_ins_dep_ix_sth();
- my $set_tran_display_xref_sth = $self->get_set_transcript_display_xref_sth();
- my $ins_object_xref_sth = $self->get_ins_object_xref_sth();
+ my $ins_xref_sth = $self->get_ins_xref_sth($dbi);
+ my $ins_dep_ix_sth = $self->get_ins_dep_ix_sth($dbi);
+ my $set_tran_display_xref_sth = $self->get_set_transcript_display_xref_sth($dbi);
+ my $ins_object_xref_sth = $self->get_ins_object_xref_sth($dbi);
if ($gene_id_to_stable_id->{$gene_id} =~ /LRG/) { return; }
@@ -769,10 +780,10 @@ sub set_transcript_display_xrefs{
#################################################
sub get_dbentrie_sth{
my $self = shift;
+ my $dbi = shift;
- if(!defined($self->{'_dbentrie_sth'})){
- my $sql =(<<"SQ1");
+ my $sql =(<<"SQ1");
SELECT x.label, x.xref_id, ox.object_xref_id, s.priority
FROM xref x, object_xref ox, source s
WHERE x.xref_id = ox.xref_id AND
@@ -782,9 +793,8 @@ SELECT x.label, x.xref_id, ox.object_xref_id, s.priority
ox.ensembl_id = ? AND
ox.ensembl_object_type = ?
SQ1
- $self->{'_dbentrie_sth'} = $self->xref->dbc->prepare($sql);
- }
- return $self->{'_dbentrie_sth'};
+ my $sth = $dbi->prepare($sql);
+ return $sth;
}
#################################################
@@ -794,10 +804,10 @@ SQ1
#################################################
sub get_dbentrie_with_desc_sth{
my $self = shift;
+ my $dbi = shift;
- if(!defined($self->{'_dbentrie_desc_sth'})){
- my $sql =(<<"SQD");
+ my $sql =(<<"SQD");
SELECT x.label, x.xref_id, ox.object_xref_id, s.priority, x.description
FROM xref x, object_xref ox, source s
WHERE x.xref_id = ox.xref_id AND
@@ -807,9 +817,8 @@ SELECT x.label, x.xref_id, ox.object_xref_id, s.priority, x.description
ox.ensembl_id = ? AND
ox.ensembl_object_type = ?
SQD
- $self->{'_dbentrie_desc_sth'} = $self->xref->dbc->prepare($sql);
- }
- return $self->{'_dbentrie_desc_sth'};
+ my $sth = $dbi->prepare($sql);
+ return $sth;
}
#################################################
@@ -818,9 +827,9 @@ SQD
#################################################
sub get_best_identity_sth{
my $self = shift;
+ my $dbi = shift;
- if(!defined($self->{'_best_identity_sth'})){
- my $sql =(<<"SQD");
+ my $sql =(<<"SQD");
SELECT x.xref_id, CASE WHEN ix.query_identity >= ix.target_identity
THEN ix.query_identity ELSE ix.target_identity END as best_identity
FROM xref x, object_xref ox, identity_xref ix, source s
@@ -829,9 +838,8 @@ WHERE x.xref_id = ox.xref_id AND x.source_id = s.source_id
AND ox.ox_status = 'DUMP_OUT' AND ox.ensembl_id = ?
AND ox.ensembl_object_type = ? order by best_identity DESC
SQD
- $self->{'_best_identity_sth'} = $self->xref->dbc->prepare($sql);
- }
- return $self->{'_best_identity_sth'};
+ my $sth = $dbi->prepare($sql);
+ return $sth;
}
@@ -842,10 +850,9 @@ SQD
#################################################
sub get_set_transcript_display_xref_sth {
my $self = shift;
- if(!defined($self->{'_set_tran_display'})){
- $self->{'_set_tran_display'} = $self->xref->dbc->prepare('UPDATE transcript_stable_id SET display_xref_id =? where internal_id = ?');
- }
- return $self->{'_set_tran_display'}
+ my $dbi = shift;
+ my $sth = $dbi->prepare('UPDATE transcript_stable_id SET display_xref_id =? where internal_id = ?');
+ return $sth;
}
@@ -856,10 +863,9 @@ sub get_set_transcript_display_xref_sth {
#################################################
sub get_set_gene_display_xref_sth {
my $self = shift;
- if(!defined($self->{'_set_gene_display'})){
- $self->{'_set_gene_display'} = $self->xref->dbc->prepare('UPDATE gene_stable_id SET display_xref_id =? where internal_id = ?');
- }
- return $self->{'_set_gene_display'}
+ my $dbi = shift;
+ my $sth = $dbi->prepare('UPDATE gene_stable_id SET display_xref_id =? where internal_id = ?');
+ return $sth;
}
@@ -868,13 +874,11 @@ sub get_set_gene_display_xref_sth {
###############################################
sub get_ins_xref_sth{
my $self= shift;
+ my $dbi = shift;
- if(!defined($self->{'_ins_xref_sth'})){
- my $sql = "insert ignore into xref (xref_id, source_id, accession, label, version, species_id, info_type, info_text, description) values (?, ?, ?, ?, 0, ".$self->species_id.", 'MISC', ?, ? )";
- print "_ins_xref sql is:-\n$sql\n";
- $self->{'_ins_xref_sth'} = $self->xref->dbc->prepare($sql);
- }
- return $self->{'_ins_xref_sth'};
+ my $sql = "insert ignore into xref (xref_id, source_id, accession, label, version, species_id, info_type, info_text, description) values (?, ?, ?, ?, 0, ".$self->species_id.", 'MISC', ?, ? )";
+ my $sth = $dbi->prepare($sql);
+ return $sth;
}
@@ -883,12 +887,11 @@ sub get_ins_xref_sth{
#################################################
sub get_ins_dep_ix_sth{
my $self= shift;
+ my $dbi = shift;
- if(!defined($self->{'_ins_identity_sth'})){
- my $sql = "insert into identity_xref (object_xref_id, query_identity, target_identity) values(?, ?, ?)";
- $self->{'_ins_identity_sth'} = $self->xref->dbc->prepare($sql);
- }
- return $self->{'_ins_identity_sth'};
+ my $sql = "insert into identity_xref (object_xref_id, query_identity, target_identity) values(?, ?, ?)";
+ my $sth = $dbi->prepare($sql);
+ return $sth;
}
###############################################
@@ -896,34 +899,34 @@ sub get_ins_dep_ix_sth{
###############################################
sub get_ins_object_xref_sth{
my $self= shift;
+ my $dbi = shift;
- if(!defined($self->{'_ins_object_xref_sth'})){
- my $sql = "insert into object_xref (object_xref_id, ensembl_id, ensembl_object_type, xref_id, linkage_type, ox_status, unused_priority) values (?, ?, ?, ?, 'MISC', 'DUMP_OUT', ?)";
- $self->{'_ins_object_xref_sth'} = $self->xref->dbc->prepare($sql);
- }
- return $self->{'_ins_object_xref_sth'};
+ my $sql = "insert into object_xref (object_xref_id, ensembl_id, ensembl_object_type, xref_id, linkage_type, ox_status, unused_priority) values (?, ?, ?, ?, 'MISC', 'DUMP_OUT', ?)";
+ my $sth = $dbi->prepare($sql);
+ return $sth;
}
sub find_max_ids{
my $self = shift;
+ my $dbi = shift;
my ($max_object_xref_id, $max_object_xref_id2, $max_xref_id);
- my $sth = $self->xref->dbc->prepare("SELECT MAX(object_xref_id) FROM object_xref");
+ my $sth = $dbi->prepare("SELECT MAX(object_xref_id) FROM object_xref");
$sth->execute();
$sth->bind_columns(\$max_object_xref_id);
$sth->fetch;
- $sth = $self->xref->dbc->prepare("SELECT MAX(object_xref_id) FROM identity_xref");
+ $sth = $dbi->prepare("SELECT MAX(object_xref_id) FROM identity_xref");
$sth->execute();
$sth->bind_columns(\$max_object_xref_id2);
$sth->fetch;
- $sth = $self->xref->dbc->prepare("SELECT MAX(xref_id) FROM xref");
+ $sth = $dbi->prepare("SELECT MAX(xref_id) FROM xref");
$sth->execute();
$sth->bind_columns(\$max_xref_id);
$sth->fetch;
@@ -943,6 +946,7 @@ sub get_tran_to_vega_data{
my $ref_synonym = $arg_ref->{synonym};
my $ref_xref_added = $arg_ref->{xref_added};
my $clone_based_vega_transcript_id = $arg_ref->{cbvt};
+ my $dbi = $arg_ref->{dbi};
my %tran_to_vega_ext;
my %tran_to_vega_name;
@@ -950,7 +954,7 @@ sub get_tran_to_vega_data{
my $dbname = $self->get_official_name();
- my $dbentrie_sth = $self->get_dbentrie_sth();
+ my $dbentrie_sth = $self->get_dbentrie_sth($dbi);
my ($display, $xref_id, $object_xref_id, $level);
foreach my $tran_id ( @{$ref_gene_to_transcripts->{$gene_id}} ){
@@ -1013,7 +1017,7 @@ sub get_tran_to_vega_data{
sub get_synonyms{
- my ($self, $synonym) = @_;
+ my ($self, $synonym, $dbi) = @_;
my $dbname = $self->get_official_name();
@@ -1025,7 +1029,7 @@ SELECT es.synonym, x.label
s.name = '$dbname'
SYN
- my $sth = $self->xref->dbc->prepare($syn_sql);
+ my $sth = $dbi->prepare($syn_sql);
$sth->execute();
my ($syn, $name);
$sth->bind_columns(\$syn,\$name);
@@ -1038,7 +1042,7 @@ SYN
sub get_display_label_data{
# my ($self, $label_to_id, $label_to_desc) = @_;
- my ($self, $label_to_desc) = @_;
+ my ($self, $label_to_desc, $dbi) = @_;
my $dbname = $self->get_official_name();
@@ -1050,7 +1054,7 @@ SELECT x.accession, sy.synonym, x.description
so.name like '$dbname'
GD1
- my $gd1_sth = $self->xref->dbc->prepare($gd1_sql);
+ my $gd1_sth = $dbi->prepare($gd1_sql);
$gd1_sth->execute();
my ($display_label, $acc, $syn, $desc);
@@ -1071,7 +1075,7 @@ SELECT x.accession, x.label, x.description
s.name like '$dbname'
GD2
- my $gd2_sth = $self->xref->dbc->prepare($gd2_sql);
+ my $gd2_sth = $dbi->prepare($gd2_sql);
$gd2_sth->execute();
$gd2_sth->bind_columns(\$acc,\$display_label, \$desc);
@@ -1106,10 +1110,11 @@ sub find_from_other_sources{
my $tran_source = $ref_args->{tran_source};
my $gene_id = $ref_args->{gene_id};
my $display_label_to_desc = $ref_args->{label_to_desc};
+ my $dbi = $ref_args->{dbi};
my %ignore_object = %{$ignore_object};
my ($gene_symbol, $gene_symbol_xref_id);
- my $dbentrie_sth = $self->get_dbentrie_with_desc_sth();
+ my $dbentrie_sth = $self->get_dbentrie_with_desc_sth($dbi);
my $other_name_num = $self->get_other_name_hash();
my ($display, $xref_id, $object_xref_id, $level, $desc);
@@ -1153,18 +1158,16 @@ sub find_from_other_sources{
#
sub get_delete_odn_sth{
- my ($self) = @_;
+ my ($self, $dbi) = @_;
- if(!defined($self->{_delete_odn_sth})){
- $self->{_delete_odn_sth} = $self->xref->dbc->prepare('UPDATE object_xref SET ox_status = "MULTI_DELETE" where object_xref_id = ?');
- }
- return $self->{_delete_odn_sth};
+ my $sth = $dbi->prepare('UPDATE object_xref SET ox_status = "MULTI_DELETE" where object_xref_id = ?');
+ return $sth;
}
sub set_the_best_odns{
- my ($self, $odn, $ref_list, $ref_list_ox, $ref_xref_id_to_display) = @_;
+ my ($self, $odn, $ref_list, $ref_list_ox, $ref_xref_id_to_display, $dbi) = @_;
- my $delete_odn_sth = $self->get_delete_odn_sth();
+ my $delete_odn_sth = $self->get_delete_odn_sth($dbi);
my %ODN = %$odn;
my $gene_symbol = undef;
@@ -1194,10 +1197,9 @@ sub set_the_best_odns{
# Get the vega xref_id for a symbol
#
sub get_find_odn_sth{
- my ($self, $dbname) = @_;
+ my ($self, $dbname, $dbi) = @_;
- if(!defined($self->{'_find_odn_sth'})){
- my $sql=(<<"SQ5");
+ my $sql=(<<"SQ5");
SELECT x.xref_id
FROM xref x, source s, object_xref ox
WHERE ox.xref_id = x.xref_id AND
@@ -1207,18 +1209,15 @@ SELECT x.xref_id
s.priority_description like "vega" AND
ox.ox_status ="DUMP_OUT"
SQ5
- print $sql."\n";
-
- $self->{'_find_odn_sth'} = $self->xref->dbc->prepare($sql);
- }
- return $self->{'_find_odn_sth'};
+ my $sth = $dbi->prepare($sql);
+ return $sth;
}
sub get_xref_id_for_gene_symbol{
- my ($self, $gene_symbol, $dbname) = @_;
+ my ($self, $gene_symbol, $dbname, $dbi) = @_;
my $xref_id = undef;
- my $find_odn_sth = $self->get_find_odn_sth($dbname);
+ my $find_odn_sth = $self->get_find_odn_sth($dbname, $dbi);
$find_odn_sth->execute($gene_symbol);
$find_odn_sth->bind_columns(\$xref_id);
@@ -1234,9 +1233,9 @@ sub get_xref_id_for_gene_symbol{
sub get_lrg_find_sth{
my $self = shift;
+ my $dbi = shift;
- if(!defined($self->{'_lrg_find_sth'})){
- my $sql=(<<'SQ2');
+ my $sql=(<<'SQ2');
SELECT x.label, x.xref_id, ox.object_xref_id, s.priority
FROM xref x, object_xref ox, source s
WHERE x.xref_id = ox.xref_id AND
@@ -1245,27 +1244,24 @@ SELECT x.label, x.xref_id, ox.object_xref_id, s.priority
ox.ensembl_id = ? AND
ox.ensembl_object_type = ?
SQ2
- $self->{'_lrg_find_sth'} =$self->xref->dbc->prepare($sql);
- }
- return $self->{'_lrg_find_sth'};
+ my $sth = $dbi->prepare($sql);
+ return $sth;
}
sub get_lrg_set_status_sth{
my $self = shift;
+ my $dbi = shift;
- if(!defined($self->{'_lrg_set_status_sth'})){
- $self->{'_lrg_set_status_sth'} =
- $self->xref->dbc->prepare("update object_xref set ox_status = 'NO_DISPLAY' where object_xref_id = ?");
- }
- return $self->{'_lrg_set_status_sth'};
+ my $sth = $dbi->prepare("update object_xref set ox_status = 'NO_DISPLAY' where object_xref_id = ?");
+ return $sth;
}
sub get_lrg_to_hgnc_sth{
my $self = shift;
+ my $dbi = shift;
- if(!defined($self->{'_lrg_to_hgnc_sth'})){
- my $sql=(<<'SQ4');
+ my $sql=(<<'SQ4');
SELECT x.xref_id, s.priority
FROM xref x,source s, object_xref ox
WHERE x.xref_id = ox.xref_id AND
@@ -1275,21 +1271,20 @@ SELECT x.xref_id, s.priority
ox.ox_status = 'DUMP_OUT'
ORDER BY s.priority
SQ4
- $self->{'_lrg_to_hgnc_sth'} = $self->xref->dbc->prepare($sql);
- }
- return $self->{'_lrg_to_hgnc_sth'};
+ my $sth = $dbi->prepare($sql);
+ return $sth;
}
sub find_lrg_hgnc{
- my ($self, $gene_id) =@_;
+ my ($self, $gene_id, $dbi) =@_;
my $gene_symbol;
my $gene_symbol_xref_id;
my $is_lrg = 0;
- my $lrg_find_sth = $self->get_lrg_find_sth();
- my $lrg_set_status_sth = $self->get_lrg_set_status_sth();
- my $lrg_to_hgnc_sth = $self->get_lrg_to_hgnc_sth();
+ my $lrg_find_sth = $self->get_lrg_find_sth($dbi);
+ my $lrg_set_status_sth = $self->get_lrg_set_status_sth($dbi);
+ my $lrg_to_hgnc_sth = $self->get_lrg_to_hgnc_sth($dbi);
# look for LRG_HGNC_notransfer, if found then find HGNC equiv and set to this
# print "LRG FOUND with no HGNC, should have gotten this via the alt allele table?? gene_id = $gene_id\n";
@@ -1369,6 +1364,7 @@ sub get_clone_name{
#
sub get_new_dbname_sources{
my $self = shift;
+ my $dbi = shift;
my %dbname_to_source_id;
@@ -1388,7 +1384,7 @@ EntrezGene_trans_name);
push @list, $dbname."_trans_name";
push @list, $dbname;
- my $sth = $self->xref->dbc->prepare("select source_id from source where name like ?");
+ my $sth = $dbi->prepare("select source_id from source where name like ?");
my $source_error = 0;
foreach my $source (@list){
@@ -1414,7 +1410,7 @@ EntrezGene_trans_name);
# my $list = "$odn_tran_id, $clone_based_vega_gene_id, $clone_based_ensembl_gene_id, $clone_based_ensembl_tran_id, $rfam_tran_id, $rfam_gene_id, $mirbase_tran_id, $mirbase_gene_id";
sub delete_old_data{
- my ($self, $dbname_to_source_id) = @_;
+ my ($self, $dbname_to_source_id, $dbi) = @_;
my $dbname = $self->get_official_name();
@@ -1443,7 +1439,7 @@ DELETE s
x.source_id in ( $list );
DE1
- my $sth = $self->xref->dbc->prepare($sql);
+ my $sth = $dbi->prepare($sql);
$sth->execute();
@@ -1454,7 +1450,7 @@ DELETE i
x.xref_id = o.xref_id AND
x.source_id in ( $list )
DE2
- $sth = $self->xref->dbc->prepare($del_identity_sql);
+ $sth = $dbi->prepare($del_identity_sql);
$sth->execute();
my $del_ox_sql = (<<"DE3");
@@ -1463,18 +1459,18 @@ DELETE o
WHERE x.xref_id = o.xref_id AND
x.source_id in ( $list )
DE3
- $sth = $self->xref->dbc->prepare($del_ox_sql);
+ $sth = $dbi->prepare($del_ox_sql);
$sth->execute();
my $del_x_sql = "delete x from xref x where x.source_id in ( $list )";
- $sth = $self->xref->dbc->prepare($del_x_sql);
+ $sth = $dbi->prepare($del_x_sql);
$sth->execute();
return;
}
sub clean_up_vega_tran_entries{
- my ($self, $clone_based_vega_tran_id) = @_;
+ my ($self, $clone_based_vega_tran_id, $dbi) = @_;
my $del_synonym_sql = (<<"CUV1"); # original ones added have info type of "DIRECT"
DELETE s
@@ -1485,7 +1481,7 @@ DELETE s
CUV1
- my $sth = $self->xref->dbc->prepare($del_synonym_sql);
+ my $sth = $dbi->prepare($del_synonym_sql);
$sth->execute();
my $del_x_sql = (<<"CUV2"); # original ones added have info type of "DIRECT"
@@ -1495,7 +1491,7 @@ DELETE x
x.info_type = 'MISC'
CUV2
- $sth = $self->xref->dbc->prepare($del_x_sql);
+ $sth = $dbi->prepare($del_x_sql);
$sth->execute();
return;
}
@@ -1503,8 +1499,9 @@ CUV2
sub reset_display_xrefs{
my $self = shift;
+ my $dbi = shift;
- my $sth = $self->xref->dbc->prepare("update transcript_stable_id set display_xref_id = null");
+ my $sth = $dbi->prepare("update transcript_stable_id set display_xref_id = null");
$sth->execute;
$sth = $self->xref->dbc->prepare("UPDATE gene_stable_id SET display_xref_id = null, desc_set =0");
diff --git a/misc-scripts/xref_mapping/XrefMapper/ProcessMappings.pm b/misc-scripts/xref_mapping/XrefMapper/ProcessMappings.pm
index ca738ddc4b..13b0b14036 100644
--- a/misc-scripts/xref_mapping/XrefMapper/ProcessMappings.pm
+++ b/misc-scripts/xref_mapping/XrefMapper/ProcessMappings.pm
@@ -71,9 +71,11 @@ sub process_mappings {
my %query_cutoff;
my %target_cutoff;
my ($job_id, $percent_query_cutoff, $percent_target_cutoff);
- my $sth = $self->xref->dbc->prepare("select job_id, percent_query_cutoff, percent_target_cutoff from mapping");
+ my $dbi = $self->xref->dbc();
+ my $sth = $dbi->prepare("select job_id, percent_query_cutoff, percent_target_cutoff from mapping");
$sth->execute();
$sth->bind_columns(\$job_id, \$percent_query_cutoff, \$percent_target_cutoff);
+ my $object_xref_id;
while($sth->fetch){
$query_cutoff{$job_id} = $percent_query_cutoff;
@@ -83,7 +85,7 @@ sub process_mappings {
my ($root_dir, $map, $status, $out, $err, $array_number);
my ($map_file, $out_file, $err_file);
- my $map_sth = $self->xref->dbc->prepare("select root_dir, map_file, status, out_file, err_file, array_number, job_id from mapping_jobs");
+ my $map_sth = $dbi->prepare("select root_dir, map_file, status, out_file, err_file, array_number, job_id from mapping_jobs");
$map_sth->execute();
$map_sth->bind_columns(\$root_dir, \$map, \$status, \$out, \$err, \$array_number, \$job_id);
my $already_processed_count = 0;
@@ -91,7 +93,7 @@ sub process_mappings {
my $error_count = 0;
my $empty_count = 0;
- my $stat_sth = $self->xref->dbc->prepare("update mapping_jobs set status = ? where job_id = ? and array_number = ?");
+ my $stat_sth = $dbi->prepare("update mapping_jobs set status = ? where job_id = ? and array_number = ?");
while($map_sth->fetch()){
my $err_file = $root_dir."/".$err;
@@ -119,7 +121,7 @@ sub process_mappings {
}
else{ #err file checks out so process the mapping file.
if(-e $map_file){
- my $count = $self->process_map_file($map_file, $query_cutoff{$job_id}, $target_cutoff{$job_id}, $job_id, $array_number);
+ my $count = $self->process_map_file($map_file, $query_cutoff{$job_id}, $target_cutoff{$job_id}, $job_id, $array_number, $dbi);
if( $count > 0){
$processed_count++;
$stat_sth->execute('SUCCESS',$job_id, $array_number);
@@ -149,7 +151,7 @@ sub process_mappings {
print "already processed = $already_processed_count, processed = $processed_count, errors = $error_count, empty = $empty_count\n" if($self->verbose);
if(!$error_count){
- my $sth = $self->xref->dbc->prepare("insert into process_status (status, date) values('mapping_processed',now())");
+ my $sth = $dbi->prepare("insert into process_status (status, date) values('mapping_processed',now())");
$sth->execute();
$sth->finish;
}
@@ -160,13 +162,13 @@ sub process_mappings {
#return number of lines parsed if succesfull. -1 for fail
sub process_map_file{
- my ($self, $map_file, $query_cutoff, $target_cutoff, $job_id, $array_number) = @_;
+ my ($self, $map_file, $query_cutoff, $target_cutoff, $job_id, $array_number, $dbi) = @_;
my $ret = 1;
my $ensembl_type = "Translation";
- if($map_file =~ /_dna_/){
+ if($map_file =~ /dna_/){
$ensembl_type = "Transcript";
}
@@ -179,14 +181,14 @@ sub process_map_file{
my $total_lines = 0;
my $root_dir = $self->core->dir;
- my $ins_go_sth = $self->xref->dbc->prepare("insert ignore into go_xref (object_xref_id, linkage_type, source_xref_id) values(?,?,?)");
- my $dep_sth = $self->xref->dbc->prepare("select dependent_xref_id, linkage_annotation from dependent_xref where master_xref_id = ?");
- my $start_sth = $self->xref->dbc->prepare("update mapping_jobs set object_xref_start = ? where job_id = ? and array_number = ?");
- my $end_sth = $self->xref->dbc->prepare("update mapping_jobs set object_xref_end = ? where job_id = ? and array_number = ?");
-# my $update_dependent_xref_sth = $self->xref->dbc->prepare("update dependent_xref set object_xref_id = ? where master_xref_id = ? and dependent_xref_id =?");
+ my $ins_go_sth = $dbi->prepare("insert ignore into go_xref (object_xref_id, linkage_type, source_xref_id) values(?,?,?)");
+ my $dep_sth = $dbi->prepare("select dependent_xref_id, linkage_annotation from dependent_xref where master_xref_id = ?");
+ my $start_sth = $dbi->prepare("update mapping_jobs set object_xref_start = ? where job_id = ? and array_number = ?");
+ my $end_sth = $dbi->prepare("update mapping_jobs set object_xref_end = ? where job_id = ? and array_number = ?");
+# my $update_dependent_xref_sth = $dbi->prepare("update dependent_xref set object_xref_id = ? where master_xref_id = ? and dependent_xref_id =?");
my $object_xref_id;
- my $sth = $self->xref->dbc->prepare("select max(object_xref_id) from object_xref");
+ my $sth = $dbi->prepare("select max(object_xref_id) from object_xref");
$sth->execute();
$sth->bind_columns(\$object_xref_id);
$sth->fetch();
@@ -196,19 +198,21 @@ sub process_map_file{
}
- my $object_xref_sth = $self->xref->dbc->prepare("insert into object_xref (object_xref_id, ensembl_id,ensembl_object_type, xref_id, linkage_type, ox_status ) values (?, ?, ?, ?, ?, ?)");
- my $object_xref_sth2 = $self->xref->dbc->prepare("insert into object_xref (object_xref_id, ensembl_id,ensembl_object_type, xref_id, linkage_type, ox_status, master_xref_id ) values (?, ?, ?, ?, ?, ?, ?)");
+ my $object_xref_sth = $dbi->prepare("insert into object_xref (ensembl_id,ensembl_object_type, xref_id, linkage_type, ox_status ) values (?, ?, ?, ?, ?)");
+ my $object_xref_sth2 = $dbi->prepare("insert into object_xref (ensembl_id,ensembl_object_type, xref_id, linkage_type, ox_status, master_xref_id ) values (?, ?, ?, ?, ?, ?)");
+ my $get_object_xref_id_sth = $dbi->prepare("select object_xref_id from object_xref where ensembl_id = ? and ensembl_object_type = ? and xref_id = ? and linkage_type = ? and ox_status = ?");
+ my $get_object_xref_id_master_sth = $dbi->prepare("select object_xref_id from object_xref where ensembl_id = ? and ensembl_object_type = ? and xref_id = ? and linkage_type = ? and ox_status = ? and master_xref_id = ?");
local $object_xref_sth->{RaiseError}; #catch duplicates
local $object_xref_sth->{PrintError}; # cut down on error messages
local $object_xref_sth2->{RaiseError}; #catch duplicates
local $object_xref_sth2->{PrintError}; # cut down on error messages
- my $identity_xref_sth = $self->xref->dbc->prepare("insert ignore into identity_xref (object_xref_id, query_identity, target_identity, hit_start, hit_end, translation_start, translation_end, cigar_line, score ) values (?, ?, ?, ?, ?, ?, ?, ?, ?)");
+ my $identity_xref_sth = $dbi->prepare("insert ignore into identity_xref (object_xref_id, query_identity, target_identity, hit_start, hit_end, translation_start, translation_end, cigar_line, score ) values (?, ?, ?, ?, ?, ?, ?, ?, ?)");
- my $ins_dep_ix_sth = $self->xref->dbc->prepare("insert ignore into identity_xref (object_xref_id, query_identity, target_identity) values(?, ?, ?)");
+ my $ins_dep_ix_sth = $dbi->prepare("insert ignore into identity_xref (object_xref_id, query_identity, target_identity) values(?, ?, ?)");
- my $source_name_sth = $self->xref->dbc->prepare("select s.name from xref x join source s using(source_id) where x.xref_id = ?");
- my $biotype_sth = $self->xref->dbc->prepare("select biotype from transcript_stable_id where internal_id = ?");
+ my $source_name_sth = $dbi->prepare("select s.name from xref x join source s using(source_id) where x.xref_id = ?");
+ my $biotype_sth = $dbi->prepare("select biotype from transcript_stable_id where internal_id = ?");
my $last_query_id = 0;
my $best_match_found = 0;
@@ -300,8 +304,9 @@ sub process_map_file{
$status = "FAILED_CUTOFF";
}
- $object_xref_id++;
- $object_xref_sth->execute($object_xref_id, $target_id, $ensembl_type, $query_id, 'SEQUENCE_MATCH', $status) ;
+ $object_xref_sth->execute($target_id, $ensembl_type, $query_id, 'SEQUENCE_MATCH', $status) ;
+ $get_object_xref_id_sth->execute($target_id, $ensembl_type, $query_id, 'SEQUENCE_MATCH', $status);
+ $object_xref_id = ($get_object_xref_id_sth->fetchrow_array())[0];
if($object_xref_sth->err){
my $err = $object_xref_sth->errstr;
if($err =~ /Duplicate/){
@@ -338,8 +343,9 @@ sub process_map_file{
$dep_sth->execute($master_xref_id);
$dep_sth->bind_columns(\$dep_xref_id, \$link);
while($dep_sth->fetch){
- $object_xref_id++;
- $object_xref_sth2->execute($object_xref_id, $target_id, $ensembl_type, $dep_xref_id, 'DEPENDENT', $status, $master_xref_id);
+ $object_xref_sth2->execute($target_id, $ensembl_type, $dep_xref_id, 'DEPENDENT', $status, $master_xref_id);
+ $get_object_xref_id_master_sth->execute($target_id, $ensembl_type, $dep_xref_id, 'DEPENDENT', $status, $master_xref_id);
+ $object_xref_id = ($get_object_xref_id_master_sth->fetchrow_array())[0];
if($object_xref_sth2->err){
my $err = $object_xref_sth->errstr;
if($err =~ /Duplicate/){
diff --git a/misc-scripts/xref_mapping/XrefMapper/ProcessPaired.pm b/misc-scripts/xref_mapping/XrefMapper/ProcessPaired.pm
index 8cc60f3fd2..0f64579558 100644
--- a/misc-scripts/xref_mapping/XrefMapper/ProcessPaired.pm
+++ b/misc-scripts/xref_mapping/XrefMapper/ProcessPaired.pm
@@ -53,32 +53,26 @@ sub process{
print "Process Pairs\n" if($self->verbose);
-
- my $sth = $self->xref->dbc->prepare("select MAX(object_xref_id) from object_xref");
- $sth->execute;
- my ($object_xref_id) = $sth->fetchrow_array();
- $object_xref_id++;
- $sth->finish;
-
- print "Starting at object_xref of $object_xref_id\n" if($self->verbose);
-
+ my $dbi = $self->xref->dbc;
+ my $object_xref_id;
#this query gives us transcript RefSeq_mRNA% object xrefs, and the paired RefSeq_peptide% accession as well as the translation id for the transcript
- my $transcr_obj_xrefs_sth = $self->xref->dbc->prepare("select gtt.translation_id, p.source_id, p.accession1, ix.query_identity, ix.target_identity from object_xref ox join xref x on (ox.xref_id = x.xref_id and ox.ox_status = 'DUMP_OUT') join source s on (x.source_id = s.source_id and s.name like 'RefSeq\_mRNA%') join pairs p on (x.accession = p.accession2) join gene_transcript_translation gtt on (gtt.transcript_id = ox.ensembl_id) join identity_xref ix using(object_xref_id)");
+ my $transcr_obj_xrefs_sth = $dbi->prepare("select gtt.translation_id, p.source_id, p.accession1, ix.query_identity, ix.target_identity from object_xref ox join xref x on (ox.xref_id = x.xref_id and ox.ox_status = 'DUMP_OUT') join source s on (x.source_id = s.source_id and s.name like 'RefSeq\_mRNA%') join pairs p on (x.accession = p.accession2) join gene_transcript_translation gtt on (gtt.transcript_id = ox.ensembl_id) join identity_xref ix using(object_xref_id)");
#this query is used to check if and object_xref exists for the related translation and paired RefSeq_peptide% with a status of 'DUMP_OUT'
- my $ox_translation_sth = $self->xref->dbc->prepare("select ox.object_xref_id, ox.xref_id from object_xref ox join xref x using(xref_id) where ox.ox_status in ('DUMP_OUT', 'FAILED_PRIORITY') and ox.ensembl_object_type = 'Translation' and ox.ensembl_id = ? and x.source_id = ? and x.accession = ?");
+ my $ox_translation_sth = $dbi->prepare("select ox.object_xref_id, ox.xref_id from object_xref ox join xref x using(xref_id) where ox.ox_status in ('DUMP_OUT', 'FAILED_PRIORITY') and ox.ensembl_object_type = 'Translation' and ox.ensembl_id = ? and x.source_id = ? and x.accession = ?");
- my $ox_insert_sth = $self->xref->dbc->prepare("insert into object_xref (object_xref_id, xref_id, ensembl_id, ensembl_object_type, linkage_type, ox_status) values(?, ?, ?, ?, 'INFERRED_PAIR', 'DUMP_OUT')");
+ my $ox_insert_sth = $dbi->prepare("insert into object_xref (xref_id, ensembl_id, ensembl_object_type, linkage_type, ox_status) values(?, ?, ?, 'INFERRED_PAIR', 'DUMP_OUT')");
+ my $get_object_xref_id_sth = $dbi->prepare("select object_xref_id from object_xref where xref_id = ? and ensembl_id = ? and ensembl_object_type = ? and linkage_type = 'INFERRED_PAIR' and ox_status = 'DUMP_OUT'");
- my $xref_sth = $self->xref->dbc->prepare("select xref_id from xref where accession = ? and source_id = ?");
+ my $xref_sth = $dbi->prepare("select xref_id from xref where accession = ? and source_id = ?");
- my $xref_update_sth = $self->xref->dbc->prepare("update xref set info_type = 'INFERRED_PAIR' where xref_id = ?");
- my $identity_update_sth = $self->xref->dbc->prepare("insert into identity_xref (object_xref_id, query_identity, target_identity) values(?, ?, ?)");
+ my $xref_update_sth = $dbi->prepare("update xref set info_type = 'INFERRED_PAIR' where xref_id = ?");
+ my $identity_update_sth = $dbi->prepare("insert into identity_xref (object_xref_id, query_identity, target_identity) values(?, ?, ?)");
- my $transl_object_xrefs_sth = $self->xref->dbc->prepare("select ox.object_xref_id, ox.ensembl_id, x.accession, gtt.transcript_id from gene_transcript_translation gtt join object_xref ox on (gtt.translation_id = ox.ensembl_id and ox.ensembl_object_type = 'Translation') join xref x on (ox.xref_id = x.xref_id and ox.ox_status = 'DUMP_OUT' and ox.ensembl_object_type = 'Translation') join source s on (x.source_id = s.source_id and s.name like 'RefSeq\_peptide%')");
+ my $transl_object_xrefs_sth = $dbi->prepare("select ox.object_xref_id, ox.ensembl_id, x.accession, gtt.transcript_id from gene_transcript_translation gtt join object_xref ox on (gtt.translation_id = ox.ensembl_id and ox.ensembl_object_type = 'Translation') join xref x on (ox.xref_id = x.xref_id and ox.ox_status = 'DUMP_OUT' and ox.ensembl_object_type = 'Translation') join source s on (x.source_id = s.source_id and s.name like 'RefSeq\_peptide%')");
- my $ox_mark_delete_sth = $self->xref->dbc->prepare("update object_xref set ox_status = 'MULTI_DELETE' where object_xref_id = ?");
+ my $ox_mark_delete_sth = $dbi->prepare("update object_xref set ox_status = 'MULTI_DELETE' where object_xref_id = ?");
$transcr_obj_xrefs_sth->execute();
@@ -106,7 +100,9 @@ sub process{
if (!$xref_id) {
die("Xref not found for accession $pep_accession source_id $pep_source_id");
}
- $ox_insert_sth->execute($object_xref_id, $xref_id, $translation_id, "Translation") || die "Could not insert object xref $object_xref_id: xref_id $xref_id, translation_id $translation_id" ;
+ $ox_insert_sth->execute($xref_id, $translation_id, "Translation") || die "Could not insert object xref $object_xref_id: xref_id $xref_id, translation_id $translation_id" ;
+ $get_object_xref_id_sth->execute($xref_id, $translation_id, 'Translation');
+ $object_xref_id = ($get_object_xref_id_sth->fetchrow_array())[0];
$xref_update_sth->execute($xref_id)|| die "Could not update xref_id $xref_id";
if ($query_identity && $target_identity) {
@@ -114,7 +110,6 @@ sub process{
}
$change{'translation object xrefs added'}++;
- $object_xref_id++;
$transl_object_xref_id = $object_xref_id;
}
@@ -151,7 +146,7 @@ sub process{
#change the status to 'MULTI_DELETE'
$ox_mark_delete_sth->execute($translation_object_xref_id) || die("Failed to update status to 'MULTI_DELETE for object_xref_id $translation_object_xref_id");
# Process all dependent xrefs as well
- $self->process_dependents($translation_object_xref_id, $translation_id, $transcript_id);
+ $self->process_dependents($translation_object_xref_id, $translation_id, $transcript_id, $dbi);
$change{'translation object xrefs removed'}++;
}
@@ -167,17 +162,17 @@ sub process{
}
#update process status
- my $sth_stat = $self->xref->dbc->prepare("insert into process_status (status, date) values('processed_pairs',now())");
+ my $sth_stat = $dbi->prepare("insert into process_status (status, date) values('processed_pairs',now())");
$sth_stat->execute();
$sth_stat->finish;
}
sub process_dependents {
- my ($self, $translation_object_xref_id, $translation_id, $transcript_id) = @_;
+ my ($self, $translation_object_xref_id, $translation_id, $transcript_id, $dbi) = @_;
- my $dep_tl_sth = $self->xref->dbc->prepare("select distinct dependent_ox.object_xref_id from object_xref master_ox, object_xref dependent_ox, xref dependent, xref master, dependent_xref dx where dependent.xref_id = dx.dependent_xref_id and master.xref_id = dx.master_xref_id and dependent.xref_id = dependent_ox.xref_id and master.xref_id = master_ox.xref_id and master_ox.object_xref_id = ? and dependent_ox.master_xref_id = master.xref_id and dependent_ox.ensembl_id = ? and dependent_ox.ensembl_object_type = 'Translation' and dependent_ox.ox_status = 'DUMP_OUT' ");
- my $dep_tr_sth = $self->xref->dbc->prepare("select distinct dependent_ox.object_xref_id from object_xref master_ox, object_xref dependent_ox, xref dependent, xref master, dependent_xref dx where dependent.xref_id = dx.dependent_xref_id and master.xref_id = dx.master_xref_id and dependent.xref_id = dependent_ox.xref_id and master.xref_id = master_ox.xref_id and master_ox.object_xref_id = ? and dependent_ox.master_xref_id = master.xref_id and dependent_ox.ensembl_id = ? and dependent_ox.ensembl_object_type = 'Transcript' and dependent_ox.ox_status = 'DUMP_OUT' ");
- my $ox_dx_delete_sth = $self->xref->dbc->prepare("update object_xref set ox_status = 'MULTI_DELETE' where object_xref_id = ?");
+ my $dep_tl_sth = $dbi->prepare("select distinct dependent_ox.object_xref_id from object_xref master_ox, object_xref dependent_ox, xref dependent, xref master, dependent_xref dx where dependent.xref_id = dx.dependent_xref_id and master.xref_id = dx.master_xref_id and dependent.xref_id = dependent_ox.xref_id and master.xref_id = master_ox.xref_id and master_ox.object_xref_id = ? and dependent_ox.master_xref_id = master.xref_id and dependent_ox.ensembl_id = ? and dependent_ox.ensembl_object_type = 'Translation' and dependent_ox.ox_status = 'DUMP_OUT' ");
+ my $dep_tr_sth = $dbi->prepare("select distinct dependent_ox.object_xref_id from object_xref master_ox, object_xref dependent_ox, xref dependent, xref master, dependent_xref dx where dependent.xref_id = dx.dependent_xref_id and master.xref_id = dx.master_xref_id and dependent.xref_id = dependent_ox.xref_id and master.xref_id = master_ox.xref_id and master_ox.object_xref_id = ? and dependent_ox.master_xref_id = master.xref_id and dependent_ox.ensembl_id = ? and dependent_ox.ensembl_object_type = 'Transcript' and dependent_ox.ox_status = 'DUMP_OUT' ");
+ my $ox_dx_delete_sth = $dbi->prepare("update object_xref set ox_status = 'MULTI_DELETE' where object_xref_id = ?");
my @master_object_xrefs;
my $new_master_object_xref_id;
diff --git a/misc-scripts/xref_mapping/XrefMapper/ProcessPrioritys.pm b/misc-scripts/xref_mapping/XrefMapper/ProcessPrioritys.pm
index f0a49f5ca2..fed75c4b98 100644
--- a/misc-scripts/xref_mapping/XrefMapper/ProcessPrioritys.pm
+++ b/misc-scripts/xref_mapping/XrefMapper/ProcessPrioritys.pm
@@ -56,10 +56,10 @@ sub new {
}
sub get_priority_names{
- my ($self) = @_;
+ my ($self, $dbi) = @_;
- my $psth = $self->xref->dbc->prepare("select s.priority_description, s.name from source s, xref x where x.source_id = s.source_id group by s.priority_description, s.name order by s.name") || die "prepare failed";
+ my $psth = $dbi->prepare("select s.priority_description, s.name from source s, xref x where x.source_id = s.source_id group by s.priority_description, s.name order by s.name") || die "prepare failed";
$psth->execute() || die "execute failed";
my @names;
@@ -83,15 +83,16 @@ sub get_priority_names{
sub process {
my ($self) = @_;
- my @names = $self->get_priority_names();
+ my $dbi = $self->xref->dbc;
+ my @names = $self->get_priority_names($dbi);
print "The following will be processed as priority xrefs\n" if($self->verbose);
foreach my $name (@names){
print "\t$name\n" if($self->verbose);
}
- my $update_ox_sth = $self->xref->dbc->prepare('update object_xref set ox_status = "FAILED_PRIORITY" where object_xref_id = ?');
- my $update_x_sth = $self->xref->dbc->prepare("update xref set dumped = 'NO_DUMP_ANOTHER_PRIORITY' where xref_id = ?");
+ my $update_ox_sth = $dbi->prepare('update object_xref set ox_status = "FAILED_PRIORITY" where object_xref_id = ?');
+ my $update_x_sth = $dbi->prepare("update xref set dumped = 'NO_DUMP_ANOTHER_PRIORITY' where xref_id = ?");
# 1) Set to failed all those that have no object xrefs.
@@ -104,7 +105,7 @@ sub process {
AND ox.object_xref_id is null
FSQL
- my $f_sth = $self->xref->dbc->prepare($f_sql);
+ my $f_sth = $dbi->prepare($f_sql);
foreach my $name (@names){
$f_sth->execute($name);
my ($xref_id);
@@ -129,7 +130,7 @@ FSQL
AND s.name = ?
ORDER BY x.accession DESC, s.priority ASC , identity DESC, x.xref_id DESC
NEWS
- my $new_sth = $self->xref->dbc->prepare($new_sql);
+ my $new_sth = $dbi->prepare($new_sql);
#
# Query to copy identity_xref values from one xref to another
# This is to keep alignment information event if alignment was not the best match
@@ -140,7 +141,7 @@ NEWS
WHERE object_xref_id = ?;
IDXCP
- my $idx_copy_sth = $self->xref->dbc->prepare($idx_copy_sql);
+ my $idx_copy_sth = $dbi->prepare($idx_copy_sql);
#
# Query to copy synonyms from one xref to another
@@ -151,15 +152,15 @@ IDXCP
WHERE xref_id = ?);
SYNCP
- my $syn_copy_sth = $self->xref->dbc->prepare($syn_copy_sql);
+ my $syn_copy_sth = $dbi->prepare($syn_copy_sql);
- my $best_ox_sth = $self->xref->dbc->prepare("SELECT object_xref_id FROM object_xref WHERE xref_id = ? and ensembl_object_type = ? and ensembl_id = ?");
+ my $best_ox_sth = $dbi->prepare("SELECT object_xref_id FROM object_xref WHERE xref_id = ? and ensembl_object_type = ? and ensembl_id = ?");
my $seq_score_sql = (<<SEQCP);
SELECT query_identity, target_identity, hit_start, hit_end, translation_start, translation_end, cigar_line, score, evalue
FROM identity_xref WHERE object_xref_id = ?
SEQCP
- my $seq_score_sth = $self->xref->dbc->prepare($seq_score_sql);
+ my $seq_score_sth = $dbi->prepare($seq_score_sql);
foreach my $name (@names){
@@ -200,7 +201,7 @@ SEQCP
$update_x_sth->execute($xref_id);
# Copy synonyms across if they are missing
$syn_copy_sth->execute($best_xref_id, $xref_id);
- $self->process_dependents($xref_id, $best_xref_id);
+ $self->process_dependents($xref_id, $best_xref_id, $dbi);
}
}
else{ # not DUMP_OUT
@@ -251,23 +252,23 @@ SEQCP
$idx_copy_sth->finish;
$syn_copy_sth->finish;
- my $sth = $self->xref->dbc->prepare("insert into process_status (status, date) values('prioritys_flagged',now())");
+ my $sth = $dbi->prepare("insert into process_status (status, date) values('prioritys_flagged',now())");
$sth->execute();
$sth->finish;
}
sub process_dependents{
# master xref IDs are entries for the current accession via various methods. We take dependent xrefs from the old and add to the new
- my ($self, $old_master_xref_id, $new_master_xref_id) = @_;
+ my ($self, $old_master_xref_id, $new_master_xref_id, $dbi) = @_;
- my $matching_ens_sth = $self->xref->dbc->prepare("select distinct ensembl_object_type, ensembl_id from object_xref where ox_status not in ('FAILED_CUTOFF') and xref_id = ? order by ensembl_object_type");
- my $dep_sth = $self->xref->dbc->prepare("select distinct dx.dependent_xref_id, dx.linkage_annotation, dx.linkage_source_id, ox.ensembl_object_type from dependent_xref dx, object_xref ox where ox.xref_id = dx.dependent_xref_id and ox.master_xref_id = dx.master_xref_id and dx.master_xref_id = ? order by ox.ensembl_object_type");
- my $insert_dep_x_sth = $self->xref->dbc->prepare("insert into dependent_xref(master_xref_id, dependent_xref_id, linkage_annotation, linkage_source_id) values(?, ?, ?, ?)");
- my $insert_dep_ox_sth = $self->xref->dbc->prepare("insert ignore into object_xref(master_xref_id, ensembl_object_type, ensembl_id, linkage_type, ox_status, xref_id) values(?, ?, ?, 'DEPENDENT', 'DUMP_OUT', ?)");
- my $dep_ox_sth = $self->xref->dbc->prepare("select object_xref_id from object_xref where master_xref_id = ? and ensembl_object_type = ? and ensembl_id = ? and linkage_type = 'DEPENDENT' AND ox_status = 'DUMP_OUT' and xref_id = ?");
- my $insert_dep_go_sth = $self->xref->dbc->prepare("insert ignore into go_xref values(?, ?, ?)");
- my $insert_ix_sth = $self->xref->dbc->prepare("insert ignore into identity_xref(object_xref_id, query_identity, target_identity) values(?, 100, 100)");
+ my $matching_ens_sth = $dbi->prepare("select distinct ensembl_object_type, ensembl_id from object_xref where ox_status not in ('FAILED_CUTOFF') and xref_id = ? order by ensembl_object_type");
+ my $dep_sth = $dbi->prepare("select distinct dx.dependent_xref_id, dx.linkage_annotation, dx.linkage_source_id, ox.ensembl_object_type from dependent_xref dx, object_xref ox where ox.xref_id = dx.dependent_xref_id and ox.master_xref_id = dx.master_xref_id and dx.master_xref_id = ? order by ox.ensembl_object_type");
+ my $insert_dep_x_sth = $dbi->prepare("insert into dependent_xref(master_xref_id, dependent_xref_id, linkage_annotation, linkage_source_id) values(?, ?, ?, ?)");
+ my $insert_dep_ox_sth = $dbi->prepare("insert ignore into object_xref(master_xref_id, ensembl_object_type, ensembl_id, linkage_type, ox_status, xref_id) values(?, ?, ?, 'DEPENDENT', 'DUMP_OUT', ?)");
+ my $dep_ox_sth = $dbi->prepare("select object_xref_id from object_xref where master_xref_id = ? and ensembl_object_type = ? and ensembl_id = ? and linkage_type = 'DEPENDENT' AND ox_status = 'DUMP_OUT' and xref_id = ?");
+ my $insert_dep_go_sth = $dbi->prepare("insert ignore into go_xref values(?, ?, ?)");
+ my $insert_ix_sth = $dbi->prepare("insert ignore into identity_xref(object_xref_id, query_identity, target_identity) values(?, 100, 100)");
my @master_xrefs = ($old_master_xref_id);
my $recursive = 0;
@@ -306,7 +307,7 @@ sub process_dependents{
# Remove all mappings to low priority xrefs
# Then delete any leftover identity or go xrefs of it
foreach my $ensembl_id (@{ $old_ensembl_ids{$object_type}} ) {
- $self->_detach_object_xref($xref_id, $dep_xref_id, $object_type, $ensembl_id);
+ $self->_detach_object_xref($xref_id, $dep_xref_id, $object_type, $ensembl_id, $dbi);
}
# Duplicate each dependent for the new master xref if it is the first in the chain
@@ -349,9 +350,9 @@ sub process_dependents{
# Set unimportant object_xrefs to FAILED_PRIORITY, and delete all those that remain
sub _detach_object_xref {
my $self = shift;
- my ($xref_id, $dep_xref_id, $object_type, $ensembl_id) = @_;
+ my ($xref_id, $dep_xref_id, $object_type, $ensembl_id, $dbi) = @_;
# Drop all the identity and go xrefs for the dependents of an xref
- my $remove_dep_ox_sth = $self->xref->dbc->prepare(
+ my $remove_dep_ox_sth = $dbi->prepare(
"DELETE ix, g FROM object_xref ox \
LEFT JOIN identity_xref ix ON ix.object_xref_id = ox.object_xref_id \
LEFT JOIN go_xref g ON g.object_xref_id = ox.object_xref_id \
@@ -359,12 +360,12 @@ sub _detach_object_xref {
);
# Fail the object_xrefs that did link to the deleted identity/go xrefs.
# This only updates one of potentially many, due to table contraints.
- my $update_dep_ox_sth = $self->xref->dbc->prepare(
+ my $update_dep_ox_sth = $dbi->prepare(
"UPDATE IGNORE object_xref SET ox_status = 'FAILED_PRIORITY' \
WHERE master_xref_id = ? AND ensembl_object_type = ? AND xref_id = ? AND ox_status = 'DUMP_OUT' AND ensembl_id = ?"
);
# This deletes everything left behind by the previous query.
- my $clean_dep_ox_sth = $self->xref->dbc->prepare(
+ my $clean_dep_ox_sth = $dbi->prepare(
"DELETE FROM object_xref \
WHERE master_xref_id = ? AND ensembl_object_type = ? AND xref_id = ? AND ox_status = 'DUMP_OUT' AND ensembl_id = ?"
);
diff --git a/misc-scripts/xref_mapping/XrefMapper/TestMappings.pm b/misc-scripts/xref_mapping/XrefMapper/TestMappings.pm
index 1d59472f04..bfdc7e9850 100644
--- a/misc-scripts/xref_mapping/XrefMapper/TestMappings.pm
+++ b/misc-scripts/xref_mapping/XrefMapper/TestMappings.pm
@@ -97,14 +97,15 @@ sub unlinked_entries{
my $xref_id;
my $count;
+ my $dbi = $self->xref->dbc;
- my $sth_stat = $self->xref->dbc->prepare("insert into process_status (status, date) values('tests_started',now())");
+ my $sth_stat = $dbi->prepare("insert into process_status (status, date) values('tests_started',now())");
$sth_stat->execute();
# dependent_xref and xref
my $count_sql = "select count(1) from dependent_xref d left join xref x on d.master_xref_id = x.xref_id where x.xref_id is null";
my $sql = "select distinct(d.master_xref_id) from dependent_xref d left join xref x on d.master_xref_id = x.xref_id where x.xref_id is null limit 10";
- my $sth = $self->xref->dbc->prepare($count_sql);
+ my $sth = $dbi->prepare($count_sql);
$sth->execute();
$sth->bind_columns(\$count);
$sth->fetch();
@@ -112,7 +113,7 @@ sub unlinked_entries{
if($count){
$failed = 1;
- $sth = $self->xref->dbc->prepare($sql);
+ $sth = $dbi->prepare($sql);
$sth->execute();
$sth->bind_columns(\$xref_id);
print STDERR "SQL QUERY: $sql\n";
@@ -129,7 +130,7 @@ sub unlinked_entries{
$sql = "select distinct(d.dependent_xref_id) from dependent_xref d left join xref x on d.dependent_xref_id = x.xref_id where x.xref_id is null limit 10";
- $sth = $self->xref->dbc->prepare($count_sql);
+ $sth = $dbi->prepare($count_sql);
$sth->execute();
$sth->bind_columns(\$count);
$sth->fetch();
@@ -137,7 +138,7 @@ sub unlinked_entries{
if($count){
$failed = 1;
- $sth = $self->xref->dbc->prepare($sql);
+ $sth = $dbi->prepare($sql);
$sth->execute();
$sth->bind_columns(\$xref_id);
print STDERR "SQL QUERY: $sql\n";
@@ -151,7 +152,7 @@ sub unlinked_entries{
$sql = "select distinct(d.xref_id) from primary_xref d left join xref x on d.xref_id = x.xref_id where x.xref_id is null limit 10";
- $sth = $self->xref->dbc->prepare($count_sql);
+ $sth = $dbi->prepare($count_sql);
$sth->execute();
$sth->bind_columns(\$count);
$sth->fetch();
@@ -159,7 +160,7 @@ sub unlinked_entries{
if($count){
$failed = 1;
- $sth = $self->xref->dbc->prepare($sql);
+ $sth = $dbi->prepare($sql);
$sth->execute();
$sth->bind_columns(\$xref_id);
print STDERR "SQL QUERY: $sql\n";
@@ -174,7 +175,7 @@ sub unlinked_entries{
$sql = "select distinct(d.general_xref_id) from ".$type."_direct_xref d left join xref x on d.general_xref_id = x.xref_id where x.xref_id is null limit 10";
- $sth = $self->xref->dbc->prepare($count_sql);
+ $sth = $dbi->prepare($count_sql);
$sth->execute();
$sth->bind_columns(\$count);
$sth->fetch();
@@ -182,7 +183,7 @@ sub unlinked_entries{
if($count){
$failed = 1;
- $sth = $self->xref->dbc->prepare($sql);
+ $sth = $dbi->prepare($sql);
$sth->execute();
$sth->bind_columns(\$xref_id);
print STDERR "SQL QUERY: $sql\n";
@@ -199,7 +200,7 @@ sub unlinked_entries{
$sql = "select distinct(d.xref_id) from synonym d left join xref x on d.xref_id = x.xref_id where x.xref_id is null limit 10";
- $sth = $self->xref->dbc->prepare($count_sql);
+ $sth = $dbi->prepare($count_sql);
$sth->execute();
$sth->bind_columns(\$count);
$sth->fetch();
@@ -207,7 +208,7 @@ sub unlinked_entries{
if($count){
$failed = 1;
- $sth = $self->xref->dbc->prepare($sql);
+ $sth = $dbi->prepare($sql);
$sth->execute();
$sth->bind_columns(\$xref_id);
print STDERR "SQL QUERY: $sql\n";
@@ -222,7 +223,7 @@ sub unlinked_entries{
$sql = "select distinct(d.object_xref_id) from identity_xref d left join object_xref o on d.object_xref_id = o.object_xref_id where o.object_xref_id is null limit 10";
- $sth = $self->xref->dbc->prepare($count_sql);
+ $sth = $dbi->prepare($count_sql);
$sth->execute();
$sth->bind_columns(\$count);
$sth->fetch();
@@ -230,7 +231,7 @@ sub unlinked_entries{
if($count){
$failed = 1;
- $sth = $self->xref->dbc->prepare($sql);
+ $sth = $dbi->prepare($sql);
$sth->execute();
$sth->bind_columns(\$xref_id);
print STDERR "SQL QUERY: $sql\n";
@@ -244,7 +245,7 @@ sub unlinked_entries{
$sql = "select distinct(d.object_xref_id) from go_xref d left join object_xref o on d.object_xref_id = o.object_xref_id where o.object_xref_id is null limit 10";
- $sth = $self->xref->dbc->prepare($count_sql);
+ $sth = $dbi->prepare($count_sql);
$sth->execute();
$sth->bind_columns(\$count);
$sth->fetch();
@@ -252,7 +253,7 @@ sub unlinked_entries{
if($count){
$failed = 1;
- $sth = $self->xref->dbc->prepare($sql);
+ $sth = $dbi->prepare($sql);
$sth->execute();
$sth->bind_columns(\$xref_id);
print STDERR "SQL QUERY: $sql\n";
@@ -268,7 +269,7 @@ sub unlinked_entries{
$sql = "select distinct(d.".$type."_id) from gene_transcript_translation d left join ".$type."_stable_id x on d.".$type."_id = x.internal_id where x.internal_id is null and d.".$type."_id is not null limit 10";
- $sth = $self->xref->dbc->prepare($count_sql);
+ $sth = $dbi->prepare($count_sql);
$sth->execute();
$sth->bind_columns(\$count);
$sth->fetch();
@@ -276,7 +277,7 @@ sub unlinked_entries{
if($count){
$failed = 1;
- $sth = $self->xref->dbc->prepare($sql);
+ $sth = $dbi->prepare($sql);
$sth->execute();
$sth->bind_columns(\$xref_id);
print STDERR "SQL QUERY: $sql\n";
@@ -292,7 +293,7 @@ sub unlinked_entries{
$count_sql = "select count(1) from xref x, source s, object_xref o left join go_xref g on o.object_xref_id = g.object_xref_id where x.xref_id = o.xref_id and s.source_id = x.source_id and s.name like 'GO' and ox_status in ('DUMP_OUT') and g.object_xref_id is null";
$sql = "select distinct(o.object_xref_id) from xref x, source s, object_xref o left join go_xref g on o.object_xref_id = g.object_xref_id where x.xref_id = o.xref_id and s.source_id = x.source_id and s.name like 'GO' and ox_status in ('DUMP_OUT') and g.object_xref_id is null limit 10";
- $sth = $self->xref->dbc->prepare($count_sql);
+ $sth = $dbi->prepare($count_sql);
$sth->execute();
$sth->bind_columns(\$count);
$sth->fetch();
@@ -300,7 +301,7 @@ sub unlinked_entries{
if($count){
$failed = 1;
- $sth = $self->xref->dbc->prepare($sql);
+ $sth = $dbi->prepare($sql);
$sth->execute();
$sth->bind_columns(\$xref_id);
print STDERR "SQL QUERY: $sql\n";
@@ -311,11 +312,11 @@ sub unlinked_entries{
}
if(!$failed){
- $sth_stat = $self->xref->dbc->prepare("insert into process_status (status, date) values('tests_finished',now())");
+ $sth_stat = $dbi->prepare("insert into process_status (status, date) values('tests_finished',now())");
$sth_stat->execute();
}
else{
- $sth_stat = $self->xref->dbc->prepare("insert into process_status (status, date) values('tests_failed',now())");
+ $sth_stat = $dbi->prepare("insert into process_status (status, date) values('tests_failed',now())");
$sth_stat->execute();
}
$sth_stat->finish;
@@ -332,8 +333,9 @@ sub entry_number_check{
my %old_object_xref_count;
my %new_object_xref_count;
+ my $dbi = $self->xref->dbc;
- my $sth = $self->xref->dbc->prepare('select s.name, count(distinct x.xref_id, ensembl_id) from xref x, object_xref ox, source s where ox.xref_id = x.xref_id and x.source_id = s.source_id and ox_status = "DUMP_OUT" and s.name not like "AFFY%" group by s.name');
+ my $sth = $dbi->prepare('select s.name, count(distinct x.xref_id, ensembl_id) from xref x, object_xref ox, source s where ox.xref_id = x.xref_id and x.source_id = s.source_id and ox_status = "DUMP_OUT" and s.name not like "AFFY%" group by s.name');
$sth->execute();
my ($name, $count);
$sth->bind_columns(\$name,\$count);
@@ -380,6 +382,7 @@ sub name_change_check{
my %new_name; # $old_name{$gene_id} = HGNC_%name
my %id_to_stable_id;
+ my $dbi = $self->xref->dbc;
my $official_name = $self->mapper->get_official_name;
if(!defined($official_name)){
@@ -389,7 +392,7 @@ sub name_change_check{
my $sql = 'select x.label, gsi.internal_id, gsi.stable_id from object_xref ox, xref x, gene_stable_id gsi, source s where x.xref_id = ox.xref_id and ox.ensembl_object_type = "Gene" and gsi.internal_id = ox.ensembl_id and x.source_id = s.source_id and s.name like "'.$official_name.'_%"';
- my $sth = $self->xref->dbc->prepare($sql);
+ my $sth = $dbi->prepare($sql);
$sth->execute();
my ($name, $gene_id, $stable_id);
$sth->bind_columns(\$name,\$gene_id, \$stable_id);
@@ -408,7 +411,7 @@ sub name_change_check{
# Use synonyms as well.
my %alias;
$sql = 'select x.label, sy.synonym from xref x, synonym sy, source so where x.xref_id = sy.xref_id and so.source_id = x.source_id and so.name like "'.$official_name.'_%" ';
- $sth = $self->xref->dbc->prepare($sql);
+ $sth = $dbi->prepare($sql);
$sth->execute();
my ($syn);
$sth->bind_columns(\$name,\$syn);
@@ -419,7 +422,7 @@ sub name_change_check{
$sth->finish;
$sql = 'select x.label, sy.synonym from xref x, synonym sy, source so where x.xref_id = sy.xref_id and so.source_id = x.source_id and so.name like "EntrezGene"';
- $sth = $self->xref->dbc->prepare($sql);
+ $sth = $dbi->prepare($sql);
$sth->execute();
$sth->bind_columns(\$name,\$syn);
while($sth->fetch()){
@@ -457,12 +460,13 @@ sub name_change_check{
sub direct_stable_id_check{
my ($self) = @_;
+ my $dbi = $self->xref->dbc;
foreach my $type (qw(gene transcript translation)){
my $sql = "select s.name, count(*) from source s, xref x, ".$type."_direct_xref gdx left join ".$type."_stable_id gsi on gdx.ensembl_stable_id = gsi.stable_id where s.source_id = x.source_id and x.xref_id = gdx.general_xref_id and gsi.stable_id is null group by s.name";
- my $sth = $self->xref->dbc->prepare($sql);
+ my $sth = $dbi->prepare($sql);
$sth->execute();
my ($name, $count);
$sth->bind_columns(\$name,\$count);
diff --git a/misc-scripts/xref_mapping/XrefMapper/XrefLoader.pm b/misc-scripts/xref_mapping/XrefMapper/XrefLoader.pm
index cb7ac8b6fe..377a0e2e28 100644
--- a/misc-scripts/xref_mapping/XrefMapper/XrefLoader.pm
+++ b/misc-scripts/xref_mapping/XrefMapper/XrefLoader.pm
@@ -57,6 +57,8 @@ sub update{
my $verbose = $self->mapper->verbose;
+ my $core_dbi = $self->core->dbc;
+ my $xref_dbi = $self->xref->dbc;
#####################################
# first remove all the projections. #
@@ -64,7 +66,7 @@ sub update{
print "Deleting all PROJECTIONs from this database\n" if $verbose;
my $sql = "DELETE es FROM xref x, external_synonym es WHERE x.xref_id = es.xref_id and x.info_type = 'PROJECTION'";
- my $sth = $self->core->dbc->prepare($sql);
+ my $sth = $core_dbi->prepare($sql);
my $affected_rows = $sth->execute();
print "\tDeleted $affected_rows PROJECTED external_synonym row(s)\n" if $verbose;
@@ -73,17 +75,17 @@ DELETE ontology_xref
FROM ontology_xref, object_xref, xref
WHERE ontology_xref.object_xref_id = object_xref.object_xref_id AND object_xref.xref_id = xref.xref_id AND xref.info_type = 'PROJECTION'
SQL
- $sth = $self->core->dbc->prepare($sql);
+ $sth = $core_dbi->prepare($sql);
$affected_rows = $sth->execute();
print "\tDeleted $affected_rows PROJECTED ontology_xref row(s)\n" if $verbose;
$sql = "DELETE object_xref FROM object_xref, xref WHERE object_xref.xref_id = xref.xref_id AND xref.info_type = 'PROJECTION'";
- $sth = $self->core->dbc->prepare($sql);
+ $sth = $core_dbi->prepare($sql);
$affected_rows = $sth->execute();
print "\tDeleted $affected_rows PROJECTED object_xref row(s)\n" if $verbose;
$sql = "DELETE xref FROM xref WHERE xref.info_type = 'PROJECTION'";
- $sth = $self->core->dbc->prepare($sql);
+ $sth = $core_dbi->prepare($sql);
$affected_rows = $sth->execute();
print "\tDeleted $affected_rows PROJECTED xref row(s)\n" if $verbose;
@@ -95,7 +97,7 @@ SQL
my %name_to_external_db_id;
$sql = "select external_db_id, db_name from external_db";
- $sth = $self->core->dbc->prepare($sql);
+ $sth = $core_dbi->prepare($sql);
$sth->execute();
my ($id, $name);
$sth->bind_columns(\$id, \$name);
@@ -107,7 +109,7 @@ SQL
my %source_id_to_external_db_id;
$sql = 'select s.source_id, s.name from source s, xref x where x.source_id = s.source_id group by s.source_id'; # only get those of interest
- $sth = $self->xref->dbc->prepare($sql);
+ $sth = $xref_dbi->prepare($sql);
$sth->execute();
$sth->bind_columns(\$id, \$name);
while($sth->fetch()){
@@ -123,7 +125,7 @@ SQL
$sth->finish;
- $sth = $self->xref->dbc->prepare("update xref set dumped = null where dumped != 'NO_DUMP_ANOTHER_PRIORITY'"); # just incase this is being ran again
+ $sth = $xref_dbi->prepare("update xref set dumped = null where dumped != 'NO_DUMP_ANOTHER_PRIORITY'"); # just incase this is being ran again
$sth->execute;
$sth->finish;
@@ -135,21 +137,21 @@ SQL
# Delete the existing ones #
######################################
my ($count);
- $sth = $self->xref->dbc->prepare('select s.name, count(*) from xref x, object_xref ox, source s where ox.xref_id = x.xref_id and x.source_id = s.source_id group by s.name');
+ $sth = $xref_dbi->prepare('select s.name, count(*) from xref x, object_xref ox, source s where ox.xref_id = x.xref_id and x.source_id = s.source_id group by s.name');
$sth->execute();
$sth->bind_columns(\$name,\$count);
- my $synonym_sth = $self->core->dbc->prepare('DELETE external_synonym FROM external_synonym, xref WHERE external_synonym.xref_id = xref.xref_id AND xref.external_db_id = ?');
- my $go_sth = $self->core->dbc->prepare('DELETE ontology_xref.* FROM ontology_xref, object_xref, xref WHERE ontology_xref.object_xref_id = object_xref.object_xref_id AND object_xref.xref_id = xref.xref_id AND xref.external_db_id = ?');
- my $identity_sth = $self->core->dbc->prepare('DELETE identity_xref FROM identity_xref, object_xref, xref WHERE identity_xref.object_xref_id = object_xref.object_xref_id AND object_xref.xref_id = xref.xref_id AND xref.external_db_id = ?');
- my $object_sth = $self->core->dbc->prepare('DELETE object_xref FROM object_xref, xref WHERE object_xref.xref_id = xref.xref_id AND xref.external_db_id = ?');
- my $dependent_sth = $self->core->dbc->prepare('DELETE d FROM dependent_xref d, xref x WHERE d.dependent_xref_id = x.xref_id and x.external_db_id = ?');
- my $xref_sth = $self->core->dbc->prepare('DELETE FROM xref WHERE xref.external_db_id = ?');
- my $unmapped_sth = $self->core->dbc->prepare('DELETE FROM unmapped_object WHERE type="xref" and external_db_id = ?');
+ my $synonym_sth = $core_dbi->prepare('DELETE external_synonym FROM external_synonym, xref WHERE external_synonym.xref_id = xref.xref_id AND xref.external_db_id = ?');
+ my $go_sth = $core_dbi->prepare('DELETE ontology_xref.* FROM ontology_xref, object_xref, xref WHERE ontology_xref.object_xref_id = object_xref.object_xref_id AND object_xref.xref_id = xref.xref_id AND xref.external_db_id = ?');
+ my $identity_sth = $core_dbi->prepare('DELETE identity_xref FROM identity_xref, object_xref, xref WHERE identity_xref.object_xref_id = object_xref.object_xref_id AND object_xref.xref_id = xref.xref_id AND xref.external_db_id = ?');
+ my $object_sth = $core_dbi->prepare('DELETE object_xref FROM object_xref, xref WHERE object_xref.xref_id = xref.xref_id AND xref.external_db_id = ?');
+ my $dependent_sth = $core_dbi->prepare('DELETE d FROM dependent_xref d, xref x WHERE d.dependent_xref_id = x.xref_id and x.external_db_id = ?');
+ my $xref_sth = $core_dbi->prepare('DELETE FROM xref WHERE xref.external_db_id = ?');
+ my $unmapped_sth = $core_dbi->prepare('DELETE FROM unmapped_object WHERE type="xref" and external_db_id = ?');
- my $transaction_start_sth = $self->core->dbc->prepare('start transaction');
- my $transaction_end_sth = $self->core->dbc->prepare('commit');
+ my $transaction_start_sth = $core_dbi->prepare('start transaction');
+ my $transaction_end_sth = $core_dbi->prepare('commit');
#
# ?? Is it faster to delete them all in one go with a external_db_id in (....) ???
@@ -201,7 +203,7 @@ SQL
# Get the offsets for object_xref, xref #
##########################################
- $sth = $self->core->dbc->prepare('select MAX(xref_id) from xref');
+ $sth = $core_dbi->prepare('select MAX(xref_id) from xref');
my $xref_offset;
$sth->execute;
$sth->bind_columns(\$xref_offset);
@@ -211,7 +213,7 @@ SQL
$self->add_meta_pair("xref_offset", $xref_offset);
- $sth = $self->core->dbc->prepare('select MAX(object_xref_id) from object_xref');
+ $sth = $core_dbi->prepare('select MAX(object_xref_id) from object_xref');
my $object_xref_offset;
$sth->execute;
$sth->bind_columns(\$object_xref_offset);
@@ -253,7 +255,7 @@ SELECT x.xref_id, x.accession, x.label, x.version, x.description, x.info_text,
x.info_type = ? order by x.xref_id
DIRS
- my $seq_sth = $self->xref->dbc->prepare($seq_sql);
+ my $seq_sth = $xref_dbi->prepare($seq_sql);
###########################
@@ -271,9 +273,9 @@ SELECT x.xref_id, x.accession, x.label, x.version, x.description, x.info_text,
x.info_type = ? order by x.xref_id
DIRS
- my $dir_sth = $self->xref->dbc->prepare($dir_sql);
+ my $dir_sth = $xref_dbi->prepare($dir_sql);
-# $dependent_sth = $self->xref->dbc->prepare('select x.xref_id, x.accession, x.label, x.version, x.description, x.info_text, ox.object_xref_id, ox.ensembl_id, ox.ensembl_object_type, d.master_xref_id from xref x, object_xref ox, dependent_xref d where ox.ox_status = "DUMP_OUT" and ox.xref_id = x.xref_id and d.object_xref_id = ox.object_xref_id and x.source_id = ? and x.info_type = ? order by x.xref_id, ox.ensembl_id');
+# $dependent_sth = $xref_dbi->prepare('select x.xref_id, x.accession, x.label, x.version, x.description, x.info_text, ox.object_xref_id, ox.ensembl_id, ox.ensembl_object_type, d.master_xref_id from xref x, object_xref ox, dependent_xref d where ox.ox_status = "DUMP_OUT" and ox.xref_id = x.xref_id and d.object_xref_id = ox.object_xref_id and x.source_id = ? and x.info_type = ? order by x.xref_id, ox.ensembl_id');
my $dep_sql =(<<DSQL);
SELECT x.xref_id, x.accession, x.label, x.version, x.description, x.info_text,
@@ -286,7 +288,7 @@ SELECT x.xref_id, x.accession, x.label, x.version, x.description, x.info_text,
ORDER BY x.xref_id, ox.ensembl_id
DSQL
- $dependent_sth = $self->xref->dbc->prepare($dep_sql);
+ $dependent_sth = $xref_dbi->prepare($dep_sql);
my $go_sql =(<<GSQL);
@@ -302,7 +304,7 @@ DSQL
order by x.xref_id, ox.ensembl_id
GSQL
- $go_sth = $self->xref->dbc->prepare($go_sql);
+ $go_sth = $xref_dbi->prepare($go_sql);
my $go_count_sql = (<<GCNTSQL);
SELECT count(*)
@@ -316,13 +318,13 @@ GCNTSQL
# SQL to add data to core
#########################
- my $add_identity_xref_sth = $self->core->dbc->prepare('insert ignore into identity_xref (object_xref_id, xref_identity, ensembl_identity, xref_start, xref_end, ensembl_start, ensembl_end, cigar_line, score, evalue) values (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)');
- my $add_go_xref_sth = $self->core->dbc->prepare('insert ignore into ontology_xref (object_xref_id, source_xref_id, linkage_type) values (?, ?, ?)');
- my $add_dependent_xref_sth = $self->core->dbc->prepare('insert ignore into dependent_xref (object_xref_id, master_xref_id, dependent_xref_id) values (?, ?, ?)');
- my $add_syn_sth = $self->core->dbc->prepare('insert ignore into external_synonym (xref_id, synonym) values (?, ?)');
- my $add_release_info_sth = $self->core->dbc->prepare('update external_db set db_release = ? where external_db_id = ?');
+ my $add_identity_xref_sth = $core_dbi->prepare('insert ignore into identity_xref (object_xref_id, xref_identity, ensembl_identity, xref_start, xref_end, ensembl_start, ensembl_end, cigar_line, score, evalue) values (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)');
+ my $add_go_xref_sth = $core_dbi->prepare('insert ignore into ontology_xref (object_xref_id, source_xref_id, linkage_type) values (?, ?, ?)');
+ my $add_dependent_xref_sth = $core_dbi->prepare('insert ignore into dependent_xref (object_xref_id, master_xref_id, dependent_xref_id) values (?, ?, ?)');
+ my $add_syn_sth = $core_dbi->prepare('insert ignore into external_synonym (xref_id, synonym) values (?, ?)');
+ my $add_release_info_sth = $core_dbi->prepare('update external_db set db_release = ? where external_db_id = ?');
- $sth = $self->xref->dbc->prepare('select s.name, s.source_id, count(*), x.info_type, s.priority_description, s.source_release from xref x, object_xref ox, source s where ox.xref_id = x.xref_id and x.source_id = s.source_id and ox_status = "DUMP_OUT" group by s.name, s.source_id, x.info_type');
+ $sth = $xref_dbi->prepare('select s.name, s.source_id, count(*), x.info_type, s.priority_description, s.source_release from xref x, object_xref ox, source s where ox.xref_id = x.xref_id and x.source_id = s.source_id and ox_status = "DUMP_OUT" group by s.name, s.source_id, x.info_type');
$sth->execute();
my ($type, $source_id, $where_from, $release_info);
$sth->bind_columns(\$name,\$source_id, \$count, \$type, \$where_from, \$release_info);
@@ -342,7 +344,7 @@ GCNTSQL
my @xref_list=(); # process at end. Add synonyms and set dumped = 1;
- my $go_count_sth = $self->xref->dbc->prepare($go_count_sql);
+ my $go_count_sth = $xref_dbi->prepare($go_count_sql);
$go_count_sth->execute($source_id, $type);
my ($go_data_present) = $go_count_sth->fetchrow_array;
$go_count_sth->finish;
@@ -367,10 +369,10 @@ GCNTSQL
if($last_xref != $xref_id){
push @xref_list, $xref_id;
$count++;
- $xref_id = $self->add_xref($xref_offset, $xref_id, $ex_id, $acc, $label, $version, $desc, $type, $info || $where_from);
+ $xref_id = $self->add_xref($xref_offset, $xref_id, $ex_id, $acc, $label, $version, $desc, $type, $info || $where_from, $core_dbi);
$last_xref = $xref_id;
}
- $object_xref_id = $self->add_object_xref($object_xref_offset, $object_xref_id, $ensembl_id, $ensembl_type, ($xref_id+$xref_offset), $analysis_ids{$ensembl_type});
+ $object_xref_id = $self->add_object_xref($object_xref_offset, $object_xref_id, $ensembl_id, $ensembl_type, ($xref_id+$xref_offset), $analysis_ids{$ensembl_type}, $core_dbi);
$add_go_xref_sth->execute( ($object_xref_id+$object_xref_offset), 0, $linkage_type);
$add_identity_xref_sth->execute( ($object_xref_id+$object_xref_offset), $query_identity, $target_identity, $hit_start, $hit_end,
$translation_start, $translation_end, $cigar_line, $score, $evalue) if $translation_start;
@@ -389,10 +391,10 @@ GCNTSQL
if($last_xref != $xref_id){
push @xref_list, $xref_id;
$count++;
- $xref_id = $self->add_xref($xref_offset, $xref_id, $ex_id, $acc, $label, $version, $desc, $type, $info || $where_from);
+ $xref_id = $self->add_xref($xref_offset, $xref_id, $ex_id, $acc, $label, $version, $desc, $type, $info || $where_from, $core_dbi);
$last_xref = $xref_id;
}
- $object_xref_id = $self->add_object_xref($object_xref_offset, $object_xref_id, $ensembl_id, $ensembl_type, ($xref_id+$xref_offset), $analysis_ids{$ensembl_type});
+ $object_xref_id = $self->add_object_xref($object_xref_offset, $object_xref_id, $ensembl_id, $ensembl_type, ($xref_id+$xref_offset), $analysis_ids{$ensembl_type}, $core_dbi);
$add_identity_xref_sth->execute( ($object_xref_id+$object_xref_offset), $query_identity, $target_identity, $hit_start, $hit_end,
$translation_start, $translation_end, $cigar_line, $score, $evalue) if $translation_start;
}
@@ -414,10 +416,10 @@ GCNTSQL
if($last_xref != $xref_id) {
push @xref_list, $xref_id;
$count++;
- $xref_id = $self->add_xref($xref_offset, $xref_id, $ex_id, $acc, $label, $version, $desc, $type, $info || $where_from);
+ $xref_id = $self->add_xref($xref_offset, $xref_id, $ex_id, $acc, $label, $version, $desc, $type, $info || $where_from, $core_dbi);
$last_xref = $xref_id;
}
- $object_xref_id = $self->add_object_xref($object_xref_offset, $object_xref_id, $ensembl_id, $ensembl_type, ($xref_id+$xref_offset), $checksum_analysis_id);
+ $object_xref_id = $self->add_object_xref($object_xref_offset, $object_xref_id, $ensembl_id, $ensembl_type, ($xref_id+$xref_offset), $checksum_analysis_id, $core_dbi);
}
print "CHECKSUM $count\n" if ($verbose);
}
@@ -437,10 +439,10 @@ GCNTSQL
if($last_xref != $xref_id){
push @xref_list, $xref_id;
$count++;
- $xref_id = $self->add_xref($xref_offset, $xref_id, $ex_id, $acc, $label, $version, $desc, $type, $info || $where_from);
+ $xref_id = $self->add_xref($xref_offset, $xref_id, $ex_id, $acc, $label, $version, $desc, $type, $info || $where_from, $core_dbi);
$last_xref = $xref_id;
}
- $object_xref_id = $self->add_object_xref($object_xref_offset, $object_xref_id, $ensembl_id, $ensembl_type, ($xref_id+$xref_offset), $analysis_ids{$ensembl_type});
+ $object_xref_id = $self->add_object_xref($object_xref_offset, $object_xref_id, $ensembl_id, $ensembl_type, ($xref_id+$xref_offset), $analysis_ids{$ensembl_type}, $core_dbi);
if(defined($master_xref_id)){ # need to sort this out as all should habe one really. (interpro generates go without these!!)
$add_dependent_xref_sth->execute(($object_xref_id+$object_xref_offset), ($master_xref_id+$xref_offset), ($xref_id+$xref_offset) );
$add_go_xref_sth->execute( ($object_xref_id+$object_xref_offset), ($master_xref_id+$xref_offset), $linkage_type);
@@ -467,10 +469,10 @@ GCNTSQL
if($last_xref != $xref_id){
push @xref_list, $xref_id;
$count++;
- $xref_id = $self->add_xref($xref_offset, $xref_id, $ex_id, $acc, $label || $acc, $version, $desc, $type, $info || $where_from);
+ $xref_id = $self->add_xref($xref_offset, $xref_id, $ex_id, $acc, $label || $acc, $version, $desc, $type, $info || $where_from, $core_dbi);
}
if($last_xref != $xref_id or $last_ensembl != $ensembl_id){
- $object_xref_id = $self->add_object_xref($object_xref_offset, $object_xref_id, $ensembl_id, $ensembl_type, ($xref_id+$xref_offset), $analysis_ids{$ensembl_type});
+ $object_xref_id = $self->add_object_xref($object_xref_offset, $object_xref_id, $ensembl_id, $ensembl_type, ($xref_id+$xref_offset), $analysis_ids{$ensembl_type}, $core_dbi);
if (defined($master_xref_id)){ # need to sort this out for FlyBase since there are EMBL direct entries from the GFF and dependent xrefs from Uniprot
$add_dependent_xref_sth->execute(($object_xref_id+$object_xref_offset), ($master_xref_id+$xref_offset), ($xref_id+$xref_offset) );
}
@@ -507,10 +509,10 @@ GCNTSQL
if($last_xref != $xref_id){
push @xref_list, $xref_id;
$count++;
- $xref_id = $self->add_xref($xref_offset, $xref_id, $ex_id, $acc, $label, $version, $desc, $type, $info || $where_from);
+ $xref_id = $self->add_xref($xref_offset, $xref_id, $ex_id, $acc, $label, $version, $desc, $type, $info || $where_from, $core_dbi);
$last_xref = $xref_id;
}
- $object_xref_id = $self->add_object_xref ($object_xref_offset, $object_xref_id, $ensembl_id, $ensembl_type, ($xref_id+$xref_offset), $analysis_ids{$ensembl_type});
+ $object_xref_id = $self->add_object_xref ($object_xref_offset, $object_xref_id, $ensembl_id, $ensembl_type, ($xref_id+$xref_offset), $analysis_ids{$ensembl_type}, $core_dbi);
$add_identity_xref_sth->execute( ($object_xref_id+$object_xref_offset), $query_identity, $target_identity, $hit_start, $hit_end,
$translation_start, $translation_end, $cigar_line, $score, $evalue);
}
@@ -525,7 +527,7 @@ GCNTSQL
if(@xref_list){
my $syn_count = 0;
my $syn_sql = "select xref_id, synonym from synonym where xref_id in(".join(", ",@xref_list).")";
- my $syn_sth = $self->xref->dbc->prepare($syn_sql);
+ my $syn_sth = $xref_dbi->prepare($syn_sql);
$syn_sth->execute();
my ($xref_id, $syn);
@@ -537,7 +539,7 @@ GCNTSQL
$syn_sth->finish;
print "\tadded $syn_count synonyms\n" if($syn_count);
- my $xref_dumped_sth = $self->xref->dbc->prepare("update xref set dumped = 'MAPPED' where xref_id in (".join(", ",@xref_list).")");
+ my $xref_dumped_sth = $xref_dbi->prepare("update xref set dumped = 'MAPPED' where xref_id in (".join(", ",@xref_list).")");
$xref_dumped_sth->execute() || die "Could not set dumped status";
$xref_dumped_sth->finish;
}
@@ -563,7 +565,7 @@ GCNTSQL
my %reason_id;
# Get the cutoff values
- $sth = $self->xref->dbc->prepare("select distinct s.name, m.percent_query_cutoff, m.percent_target_cutoff from source s, source_mapping_method sm, mapping m where sm.source_id = s.source_id and sm.method = m.method");
+ $sth = $xref_dbi->prepare("select distinct s.name, m.percent_query_cutoff, m.percent_target_cutoff from source s, source_mapping_method sm, mapping m where sm.source_id = s.source_id and sm.method = m.method");
$sth->execute();
my ($source_name, $q_cut, $t_cut);
$sth->bind_columns(\$source_name, \$q_cut, \$t_cut);
@@ -592,14 +594,14 @@ GCNTSQL
foreach my $key (keys %desc_failed){
- $sth = $self->core->dbc->prepare("select unmapped_reason_id from unmapped_reason where full_description like '".$desc_failed{$key}."'");
+ $sth = $core_dbi->prepare("select unmapped_reason_id from unmapped_reason where full_description like '".$desc_failed{$key}."'");
$sth->execute();
my $failed_id=undef;
$sth->bind_columns(\$failed_id);
$sth->fetch;
$sth->finish;
if(!defined($failed_id)){
- $sth = $self->core->dbc->prepare('insert into unmapped_reason (summary_description, full_description) values("'.$summary_failed{$key}.'", "'.$desc_failed{$key}.'")');
+ $sth = $core_dbi->prepare('insert into unmapped_reason (summary_description, full_description) values("'.$summary_failed{$key}.'", "'.$desc_failed{$key}.'")');
$sth->execute();
$failed_id = $sth->{'mysql_insertid'};
$sth->finish
@@ -624,19 +626,19 @@ GCNTSQL
AND x.info_type = 'DIRECT'
DIR
- my $direct_unmapped_sth = $self->xref->dbc->prepare($sql);
+ my $direct_unmapped_sth = $xref_dbi->prepare($sql);
my ($xref_id, $acc, $version, $label, $desc, $info);
$direct_unmapped_sth->execute();
$direct_unmapped_sth->bind_columns(\$xref_id, \$acc, \$version, \$label, \$desc, \$type, \$info, \$dbname);
- my $set_unmapped_sth = $self->core->dbc->prepare("insert into unmapped_object (type, analysis_id, external_db_id, identifier, unmapped_reason_id ) values ('xref', ?, ?, ?, ?)");
+ my $set_unmapped_sth = $core_dbi->prepare("insert into unmapped_object (type, analysis_id, external_db_id, identifier, unmapped_reason_id ) values ('xref', ?, ?, ?, ?)");
my @xref_list = ();
my $analysis_id = $analysis_ids{'Transcript'}; # No real analysis here but in table it is set to not NULL
while($direct_unmapped_sth->fetch()){
my $ex_id = $name_to_external_db_id{$dbname};
if(defined($name_to_external_db_id{$dbname})){
- $xref_id = $self->add_xref($xref_offset, $xref_id, $ex_id, $acc, $label, $version, $desc, 'UNMAPPED', $info);
+ $xref_id = $self->add_xref($xref_offset, $xref_id, $ex_id, $acc, $label, $version, $desc, 'UNMAPPED', $info, $core_dbi);
$set_unmapped_sth->execute($analysis_id, $ex_id, $acc, $reason_id{"NO_STABLE_ID"});
push @xref_list, $xref_id;
}
@@ -646,7 +648,7 @@ DIR
if(@xref_list){
- my $xref_dumped_sth = $self->xref->dbc->prepare("update xref set dumped = 'UNMAPPED_NO_STABLE_ID' where xref_id in (".join(", ",@xref_list).")");
+ my $xref_dumped_sth = $xref_dbi->prepare("update xref set dumped = 'UNMAPPED_NO_STABLE_ID' where xref_id in (".join(", ",@xref_list).")");
$xref_dumped_sth->execute();
$xref_dumped_sth->finish;
}
@@ -663,7 +665,7 @@ DIR
AND x.info_type = 'MISC'
MIS
- my $misc_unmapped_sth = $self->xref->dbc->prepare($sql);
+ my $misc_unmapped_sth = $xref_dbi->prepare($sql);
$misc_unmapped_sth->execute();
$misc_unmapped_sth->bind_columns(\$xref_id, \$acc, \$version, \$label, \$desc, \$type, \$info, \$dbname);
@@ -672,7 +674,7 @@ MIS
while($misc_unmapped_sth->fetch()){
my $ex_id = $name_to_external_db_id{$dbname};
if(defined($name_to_external_db_id{$dbname})){
- $xref_id = $self->add_xref($xref_offset, $xref_id, $ex_id, $acc, $label, $version, $desc, 'UNMAPPED', $info);
+ $xref_id = $self->add_xref($xref_offset, $xref_id, $ex_id, $acc, $label, $version, $desc, 'UNMAPPED', $info, $core_dbi);
$set_unmapped_sth->execute($analysis_id, $ex_id, $acc, $reason_id{"NO_MAPPING"});
push @xref_list, $xref_id;
}
@@ -682,7 +684,7 @@ MIS
if(@xref_list){
- my $xref_dumped_sth = $self->xref->dbc->prepare("update xref set dumped = 'UNMAPPED_NO_MAPPING' where xref_id in (".join(", ",@xref_list).")");
+ my $xref_dumped_sth = $xref_dbi->prepare("update xref set dumped = 'UNMAPPED_NO_MAPPING' where xref_id in (".join(", ",@xref_list).")");
$xref_dumped_sth->execute();
$xref_dumped_sth->finish;
}
@@ -704,12 +706,12 @@ MIS
ORDER BY s.name, x.accession
DEP
- my $dep_unmapped_sth = $self->xref->dbc->prepare($sql);
+ my $dep_unmapped_sth = $xref_dbi->prepare($sql);
$dep_unmapped_sth->execute();
my $parent;
$dep_unmapped_sth->bind_columns(\$xref_id, \$acc, \$version, \$label, \$desc, \$type, \$info, \$dbname, \$parent);
- $set_unmapped_sth = $self->core->dbc->prepare("insert ignore into unmapped_object (type, analysis_id, external_db_id, identifier, unmapped_reason_id, parent ) values ('xref', ?, ?, ?, '".$reason_id{"MASTER_FAILED"}."', ?)");
+ $set_unmapped_sth = $core_dbi->prepare("insert ignore into unmapped_object (type, analysis_id, external_db_id, identifier, unmapped_reason_id, parent ) values ('xref', ?, ?, ?, '".$reason_id{"MASTER_FAILED"}."', ?)");
@xref_list = ();
my $last_acc= 0;
@@ -719,7 +721,7 @@ DEP
next;
}
if($last_acc ne $acc){
- $xref_id = $self->add_xref($xref_offset, $xref_id, $ex_id, $acc, $label||$acc, $version, $desc, 'UNMAPPED', $info);
+ $xref_id = $self->add_xref($xref_offset, $xref_id, $ex_id, $acc, $label||$acc, $version, $desc, 'UNMAPPED', $info, $core_dbi);
}
$last_acc = $acc;
$set_unmapped_sth->execute($analysis_id, $ex_id, $acc, $parent);
@@ -730,7 +732,7 @@ DEP
if(@xref_list){
- my $xref_dumped_sth = $self->xref->dbc->prepare("update xref set dumped = 'UNMAPPED_MASTER_FAILED' where xref_id in (".join(", ",@xref_list).")");
+ my $xref_dumped_sth = $xref_dbi->prepare("update xref set dumped = 'UNMAPPED_MASTER_FAILED' where xref_id in (".join(", ",@xref_list).")");
$xref_dumped_sth->execute();
$xref_dumped_sth->finish;
}
@@ -756,13 +758,13 @@ DEP
SEQ
# removed AND ox.ox_status != 'FAILED_PRIORITY'
- my $seq_unmapped_sth = $self->xref->dbc->prepare($sql);
+ my $seq_unmapped_sth = $xref_dbi->prepare($sql);
$seq_unmapped_sth->execute();
my ($ensembl_object_type, $ensembl_id, $q_id, $t_id, $seq_type, $status) ;
$seq_unmapped_sth->bind_columns(\$xref_id, \$acc, \$version, \$label, \$desc, \$type, \$info, \$dbname, \$seq_type, \$ensembl_object_type, \$ensembl_id, \$q_id, \$t_id,\$status);
- my $set_unmapped_no_sth = $self->core->dbc->prepare("insert into unmapped_object (type, analysis_id, external_db_id, identifier, unmapped_reason_id, ensembl_object_type ) values ('xref', ?, ?, ?, '".$reason_id{"FAILED_MAP"}."', ?)");
- my $set_unmapped_failed_sth = $self->core->dbc->prepare("insert into unmapped_object (type, analysis_id, external_db_id, identifier, unmapped_reason_id, query_score, target_score, ensembl_id, ensembl_object_type ) values ('xref', ?, ?, ?, ?,?,?,?,?)");
+ my $set_unmapped_no_sth = $core_dbi->prepare("insert into unmapped_object (type, analysis_id, external_db_id, identifier, unmapped_reason_id, ensembl_object_type ) values ('xref', ?, ?, ?, '".$reason_id{"FAILED_MAP"}."', ?)");
+ my $set_unmapped_failed_sth = $core_dbi->prepare("insert into unmapped_object (type, analysis_id, external_db_id, identifier, unmapped_reason_id, query_score, target_score, ensembl_id, ensembl_object_type ) values ('xref', ?, ?, ?, ?,?,?,?,?)");
@xref_list = ();
@@ -774,7 +776,7 @@ SEQ
next;
}
if($last_xref != $xref_id){
- $xref_id = $self->add_xref($xref_offset, $xref_id, $ex_id, $acc, $label, $version, $desc, 'UNMAPPED', $info);
+ $xref_id = $self->add_xref($xref_offset, $xref_id, $ex_id, $acc, $label, $version, $desc, 'UNMAPPED', $info, $core_dbi);
}
$last_xref = $xref_id;
if(defined($ensembl_id)){
@@ -800,7 +802,7 @@ SEQ
if(@xref_list){
- my $xref_dumped_sth = $self->xref->dbc->prepare("update xref set dumped = 'UNMAPPED_NO_MAPPING' where xref_id in (".join(", ",@xref_list).")");
+ my $xref_dumped_sth = $xref_dbi->prepare("update xref set dumped = 'UNMAPPED_NO_MAPPING' where xref_id in (".join(", ",@xref_list).")");
$xref_dumped_sth->execute();
$xref_dumped_sth->finish;
}
@@ -822,11 +824,11 @@ SEQ
WEL
- my $wel_unmapped_sth = $self->xref->dbc->prepare($sql);
+ my $wel_unmapped_sth = $xref_dbi->prepare($sql);
$wel_unmapped_sth->execute();
$wel_unmapped_sth->bind_columns(\$xref_id, \$acc, \$version, \$label, \$desc, \$type, \$info, \$dbname);
- $set_unmapped_sth = $self->core->dbc->prepare("insert into unmapped_object (type, analysis_id, external_db_id, identifier, unmapped_reason_id) values ('xref', ?, ?, ?, '".$reason_id{"NO_MASTER"}."')");
+ $set_unmapped_sth = $core_dbi->prepare("insert into unmapped_object (type, analysis_id, external_db_id, identifier, unmapped_reason_id) values ('xref', ?, ?, ?, '".$reason_id{"NO_MASTER"}."')");
$analysis_id = $analysis_ids{'Transcript'}; # No real analysis here but in table it is set to not NULL
@xref_list = ();
@@ -835,7 +837,7 @@ WEL
if(!defined($ex_id)){
next;
}
- $xref_id = $self->add_xref($xref_offset, $xref_id, $ex_id, $acc, $label, $version, $desc, 'UNMAPPED', $info);
+ $xref_id = $self->add_xref($xref_offset, $xref_id, $ex_id, $acc, $label, $version, $desc, 'UNMAPPED', $info, $core_dbi);
$set_unmapped_sth->execute($analysis_id, $ex_id, $acc);
push @xref_list, $xref_id;
}
@@ -843,7 +845,7 @@ WEL
$set_unmapped_sth->finish;
if(@xref_list){
- my $xref_dumped_sth = $self->xref->dbc->prepare("update xref set dumped = 'UNMAPPED_NO_MASTER' where xref_id in (".join(", ",@xref_list).")");
+ my $xref_dumped_sth = $xref_dbi->prepare("update xref set dumped = 'UNMAPPED_NO_MASTER' where xref_id in (".join(", ",@xref_list).")");
$xref_dumped_sth->execute();
$xref_dumped_sth->finish;
}
@@ -851,7 +853,7 @@ WEL
$transaction_end_sth->execute();
- my $sth_stat = $self->xref->dbc->prepare("insert into process_status (status, date) values('core_loaded',now())");
+ my $sth_stat = $xref_dbi->prepare("insert into process_status (status, date) values('core_loaded',now())");
$sth_stat->execute();
$sth_stat->finish;
@@ -904,8 +906,7 @@ sub get_single_analysis {
sub add_xref {
- my ($self, $offset, $xref_id, $external_db_id, $dbprimary_acc, $display_label, $version, $description, $info_type, $info_text) = @_;
- my $dbc = $self->core->dbc();
+ my ($self, $offset, $xref_id, $external_db_id, $dbprimary_acc, $display_label, $version, $description, $info_type, $info_text, $dbc) = @_;
my $select_sth = $dbc->prepare("select xref_id from xref where dbprimary_acc = ? and external_db_id = ? and info_type = ? and info_text = ? and version = ?");
my $insert_sth = $dbc->prepare("insert into xref (xref_id, external_db_id, dbprimary_acc, display_label, version, description, info_type, info_text) values (?, ?, ?, ?, ?, ?, ?, ?)");
my $new_xref_id;
@@ -921,8 +922,7 @@ sub add_xref {
}
sub add_object_xref {
- my ($self, $offset, $object_xref_id, $ensembl_id, $ensembl_object_type, $xref_id, $analysis_id) = @_;
- my $dbc = $self->core->dbc();
+ my ($self, $offset, $object_xref_id, $ensembl_id, $ensembl_object_type, $xref_id, $analysis_id, $dbc) = @_;
my $select_sth = $dbc->prepare("select object_xref_id from object_xref where xref_id = ? and ensembl_object_type = ? and ensembl_id = ? and analysis_id = ?");
my $insert_sth = $dbc->prepare("insert ignore into object_xref (object_xref_id, ensembl_id, ensembl_object_type, xref_id, analysis_id) values (?, ?, ?, ?, ?)");
my $new_object_xref_id;
diff --git a/misc-scripts/xref_mapping/XrefParser/ArrayExpressParser.pm b/misc-scripts/xref_mapping/XrefParser/ArrayExpressParser.pm
index feefc4410c..ecd64bbcfe 100644
--- a/misc-scripts/xref_mapping/XrefParser/ArrayExpressParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/ArrayExpressParser.pm
@@ -38,9 +38,12 @@ sub run_script {
my ($self, $ref_arg) = @_;
my $source_id = $ref_arg->{source_id};
my $species_id = $ref_arg->{species_id};
+ my $species_name = $ref_arg->{species};
my $file = $ref_arg->{file};
my $verbose = $ref_arg->{verbose};
my $db = $ref_arg->{dba};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $file) ){
croak "Need to pass source_id, species_id and file as pairs";
@@ -73,7 +76,9 @@ sub run_script {
$user = $1;
}
- my %species_id_to_names = $self->species_id2name();
+ my %species_id_to_names = $self->species_id2name($dbi);
+ if (defined $species_name) { push @{$species_id_to_names{$species_id}}, $species_name; }
+ if (!defined $species_id_to_names{$species_id}) { next; }
my $species_id_to_names = \%species_id_to_names;
my $names = $species_id_to_names->{$species_id};
my $species_lookup = $self->_get_species($verbose);
@@ -83,7 +88,7 @@ sub run_script {
return;
}
- my $species_name = $species_id_to_names{$species_id}[0];
+ $species_name = $species_id_to_names{$species_id}[0];
#get stable_ids from core and create xrefs
@@ -140,9 +145,10 @@ sub run_script {
label => $gene_stable_id,
source_id => $source_id,
species_id => $species_id,
+ dbi => $dbi,
info_type => "DIRECT"} );
- $self->add_direct_xref( $xref_id, $gene_stable_id, 'gene', '');
+ $self->add_direct_xref( $xref_id, $gene_stable_id, 'gene', '', $dbi);
if ($xref_id) {
$xref_count++;
}
diff --git a/misc-scripts/xref_mapping/XrefParser/BaseParser.pm b/misc-scripts/xref_mapping/XrefParser/BaseParser.pm
index 56083c5473..1feaedf68e 100644
--- a/misc-scripts/xref_mapping/XrefParser/BaseParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/BaseParser.pm
@@ -31,15 +31,6 @@ use Getopt::Long;
my $base_dir = File::Spec->curdir();
-my $add_xref_sth = undef;
-my $add_object_xref_sth = undef;
-my $add_identity_xref_sth = undef;
-my %add_direct_xref_sth;
-my $add_dependent_xref_sth = undef;
-my $get_xref_sth = undef;
-my $get_object_xref_sth = undef;
-my $add_synonym_sth = undef;
-
my %xref_dependent_mapped;
@@ -139,7 +130,8 @@ sub get_filehandle
# Returns source_id or -1 if not found
#############################################
sub get_source_id_for_source_name {
- my ($self, $source_name,$priority_desc) = @_;
+ my ($self, $source_name,$priority_desc, $dbi) = @_;
+ $dbi = $self->dbi unless defined $dbi;
my $low_name = lc $source_name;
my $sql = "SELECT source_id FROM source WHERE LOWER(name)='$low_name'";
@@ -148,7 +140,7 @@ sub get_source_id_for_source_name {
$sql .= " AND LOWER(priority_description)='$low_name'";
$source_name .= " ($priority_desc)";
}
- my $sth = $self->dbi->prepare($sql);
+ my $sth = $dbi->prepare($sql);
$sth->execute();
my @row = $sth->fetchrow_array();
my $source_id;
@@ -176,12 +168,13 @@ sub get_source_id_for_source_name {
############################################################
sub get_source_ids_for_source_name_pattern {
- my ($self, $source_name) = @_;
+ my ($self, $source_name, $dbi) = @_;
+ $dbi = $self->dbi unless defined $dbi;
my $big_name = uc $source_name;
my $sql = "SELECT source_id FROM source WHERE upper(name) LIKE '%${big_name}%'";
- my $sth = $self->dbi->prepare($sql);
+ my $sth = $dbi->prepare($sql);
my @sources;
$sth->execute();
while(my @row = $sth->fetchrow_array()){
@@ -198,11 +191,12 @@ sub get_source_ids_for_source_name_pattern {
# From a source_id get the name
###############################
sub get_source_name_for_source_id {
- my ($self, $source_id) = @_;
+ my ($self, $source_id, $dbi) = @_;
+ $dbi = $self->dbi unless defined $dbi;
my $source_name;
my $sql = "SELECT name FROM source WHERE source_id= '$source_id'";
- my $sth = $self->dbi->prepare($sql);
+ my $sth = $dbi->prepare($sql);
$sth->execute();
my @row = $sth->fetchrow_array();
if (@row) {
@@ -224,11 +218,10 @@ sub get_source_name_for_source_id {
# to master_xref_id for all of source names given
#####################################################
sub get_valid_xrefs_for_dependencies{
- my ($self, $dependent_name, @reverse_ordered_source_list) = @_;
+ my ($self, $dependent_name, $dbi, @reverse_ordered_source_list) = @_;
+ $dbi = $self->dbi unless defined $dbi;
my %dependent_2_xref;
- my $dbi = $self->dbi();
-
my $sql = 'select source_id from source where LOWER(name) =?';
my $sth = $dbi->prepare($sql);
@@ -275,11 +268,10 @@ DSS
# to master_xref_id for all of source names given
#####################################################
sub get_valid_xrefs_for_direct_xrefs{
- my ($self, $direct_name, $separator) = @_;
+ my ($self, $direct_name, $separator, $dbi) = @_;
+ $dbi = $self->dbi unless defined $dbi;
my %direct_2_xref;
- my $dbi = $self->dbi;
-
my $sql = 'select source_id from source where name like ?';
my $sth = $dbi->prepare($sql);
@@ -326,8 +318,8 @@ GDS
#############################################
sub label_to_acc{
- my ($self,$source_name,$species_id) =@_;
- my $dbi = $self->dbi;
+ my ($self,$source_name,$species_id, $dbi) =@_;
+ $dbi = $self->dbi unless defined $dbi;
# First cache synonyms so we can quickly add them later
my %synonyms;
@@ -380,11 +372,11 @@ sub label_to_acc{
####################################################
sub get_valid_codes{
- my ($self,$source_name,$species_id) =@_;
+ my ($self,$source_name,$species_id, $dbi) =@_;
my %valid_codes;
my @sources;
- my $dbi = $self->dbi();
+ $dbi = $self->dbi unless defined $dbi;
my $big_name = uc $source_name;
my $sql = "select source_id from source where upper(name) like '%$big_name%'";
@@ -410,7 +402,7 @@ sub get_valid_codes{
# Upload xrefs to the database
##############################
sub upload_xref_object_graphs {
- my ($self, $rxrefs) = @_;
+ my ($self, $rxrefs, $dbi) = @_;
my $count = scalar @{$rxrefs};
if($verbose) {
@@ -431,7 +423,7 @@ sub upload_xref_object_graphs {
#################################################################################
# Start of sql needed to add xrefs, primary_xrefs, synonym, dependent_xrefs etc..
#################################################################################
- my $dbi = $self->dbi;
+ $dbi = $self->dbi unless defined $dbi;
my $xref_sth = $dbi->prepare('INSERT INTO xref (accession,version,label,description,source_id,species_id, info_type) VALUES(?,?,?,?,?,?,?)');
my $pri_insert_sth = $dbi->prepare('INSERT INTO primary_xref VALUES(?,?,?,?)');
my $pri_update_sth = $dbi->prepare('UPDATE primary_xref SET sequence=? WHERE xref_id=?');
@@ -638,19 +630,20 @@ sub upload_xref_object_graphs {
# $direct_xref is a reference to an array of hash objects.
######################################################################################
sub upload_direct_xrefs{
- my ($self, $direct_xref) = @_;
+ my ($self, $direct_xref, $dbi) = @_;
+ $dbi = $self->dbi unless defined $dbi;
for my $dr(@{$direct_xref}) {
################################################
# Find the xref_id for this accession and source
################################################
- my $general_xref_id = get_xref($dr->{ACCESSION},$dr->{SOURCE_ID},$dr->{SPECIES_ID});
+ my $general_xref_id = $self->get_xref($dr->{ACCESSION},$dr->{SOURCE_ID},$dr->{SPECIES_ID}, $dbi);
#######################################################
# If found add the direct xref else write error message
#######################################################
if ($general_xref_id){
- $self->add_direct_xref($general_xref_id, $dr->{ENSEMBL_STABLE_ID},$dr->{ENSEMBL_TYPE},$dr->{LINKAGE_XREF});
+ $self->add_direct_xref($general_xref_id, $dr->{ENSEMBL_STABLE_ID},$dr->{ENSEMBL_TYPE},$dr->{LINKAGE_XREF}, $dbi);
}
else{
print {*STDERR} 'Problem Could not find accession '.$dr->{ACCESSION}.' for source '.$dr->{SOURCE}.
@@ -668,9 +661,10 @@ sub upload_direct_xrefs{
###############################################
sub add_meta_pair {
- my ($self, $key, $value) = @_;
+ my ($self, $key, $value, $dbi) = @_;
+ $dbi = $self->dbi unless defined $dbi;
- my $sth = $self->dbi->prepare('insert into meta (meta_key, meta_value, date) values("'.$key.'", "'.$value.'", now())');
+ my $sth = $dbi->prepare('insert into meta (meta_key, meta_value, date) values("'.$key.'", "'.$value.'", now())');
$sth->execute;
$sth->finish;
return;
@@ -684,9 +678,10 @@ sub add_meta_pair {
sub get_xref_sources {
my $self = shift;
+ my $dbi = shift;
+ $dbi = $self->dbi unless defined $dbi;
my %sourcename_to_sourceid;
- my $dbi = $self->dbi;
my $sth = $dbi->prepare('SELECT name,source_id FROM source');
$sth->execute() or croak( $dbi->errstr() );
while(my @row = $sth->fetchrow_array()) {
@@ -705,9 +700,10 @@ sub get_xref_sources {
sub species_id2taxonomy {
my $self = shift;
+ my $dbi = shift;
+ $dbi = $self->dbi unless defined $dbi;
my %species_id2taxonomy;
- my $dbi = $self->dbi;
my $sth = $dbi->prepare('SELECT species_id, taxonomy_id FROM species');
$sth->execute() or croak( $dbi->errstr() );
@@ -731,10 +727,11 @@ sub species_id2taxonomy {
#########################################################################
sub species_id2name {
my $self = shift;
+ my $dbi = shift;
+ $dbi = $self->dbi unless defined $dbi;
my %species_id2name;
- my $dbi = $self->dbi;
my $sth = $dbi->prepare('SELECT species_id, name FROM species');
$sth->execute() or croak( $dbi->errstr() );
while ( my @row = $sth->fetchrow_array() ) {
@@ -771,8 +768,9 @@ sub get_xref_id {
my $source = $arg_ref->{source_id} || croak 'Need an source_id for get_xref_id';
my $species = $arg_ref->{species_id} || confess 'Need an species_id for get_xref_id';
my $error = $arg_ref->{error};
+ my $dbi = $arg_ref->{dbi};
- my $id = $self->get_xref($acc, $source, $species);
+ my $id = $self->get_xref($acc, $source, $species, $dbi);
return $id;
}
@@ -783,11 +781,11 @@ sub get_xref_id {
##################################################################
sub primary_xref_id_exists {
- my ($self, $xref_id) = @_;
+ my ($self, $xref_id, $dbi) = @_;
+ $dbi = $self->dbi unless defined $dbi;
my $exists = 0;
- my $dbi = $self->dbi;
my $sth = $dbi->prepare('SELECT xref_id FROM primary_xref WHERE xref_id=?');
$sth->execute($xref_id) or croak( $dbi->errstr() );
my @row = $sth->fetchrow_array();
@@ -802,10 +800,10 @@ sub primary_xref_id_exists {
# Get the tax id for a particular species id
############################################
sub get_taxonomy_from_species_id{
- my ($self,$species_id) = @_;
+ my ($self,$species_id, $dbi) = @_;
my %hash;
- my $dbi= $self->dbi;
+ $dbi = $self->dbi unless defined $dbi;
my $sth = $dbi->prepare("SELECT taxonomy_id FROM species WHERE species_id = $species_id");
$sth->execute() or croak( $dbi->errstr() );
while(my @row = $sth->fetchrow_array()) {
@@ -821,11 +819,11 @@ sub get_taxonomy_from_species_id{
# Only used in GOParser at the moment
################################################
sub get_direct_xref{
- my ($self,$stable_id,$type,$link) = @_;
+ my ($self,$stable_id,$type,$link, $dbi) = @_;
+ $dbi = $self->dbi unless defined $dbi;
$type = lc $type;
- my $dbi = $self->dbi;
my $sql = "select general_xref_id from ${type}_direct_xref d where ensembl_stable_id = ? and linkage_xref= ?";
my $direct_sth = $dbi->prepare($sql);
@@ -842,16 +840,14 @@ sub get_direct_xref{
# if not found return undef;
###################################################################
sub get_xref{
- my ($self,$acc,$source, $species_id) = @_;
+ my ($self,$acc,$source, $species_id, $dbi) = @_;
+ $dbi = $self->dbi unless defined $dbi;
- my $dbi = $self->dbi;
#
# If the statement handle does nt exist create it.
#
- if(!(defined $get_xref_sth) ){
- my $sql = 'select xref_id from xref where accession = ? and source_id = ? and species_id = ?';
- $get_xref_sth = $dbi->prepare($sql);
- }
+ my $sql = 'select xref_id from xref where accession = ? and source_id = ? and species_id = ?';
+ my $get_xref_sth = $dbi->prepare($sql);
#
# Find the xref_id using the sql above
@@ -868,16 +864,11 @@ sub get_xref{
# if not found return undef;
###################################################################
sub get_object_xref {
- my ($self, $xref_id, $ensembl_id, $object_type) = @_;
+ my ($self, $xref_id, $ensembl_id, $object_type, $dbi) = @_;
+ $dbi = $self->dbi unless defined $dbi;
- my $dbi = $self->dbi;
- #
- # If the statement handle does nt exist create it.
- #
- if(!(defined $get_object_xref_sth) ){
- my $sql = 'select object_xref_id from object_xref where xref_id = ? and ensembl_object_type = ? and ensembl_id = ?';
- $get_object_xref_sth = $dbi->prepare($sql);
- }
+ my $sql = 'select object_xref_id from object_xref where xref_id = ? and ensembl_object_type = ? and ensembl_id = ?';
+ my $get_object_xref_sth = $dbi->prepare($sql);
#
# Find the object_xref_id using the sql above
@@ -906,27 +897,22 @@ sub add_xref {
my $version = $arg_ref->{version} || 0;
my $info_type = $arg_ref->{info_type} || 'MISC';
my $info_text = $arg_ref->{info_text} || '';
+ my $dbi = $arg_ref->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
##################################################################
# See if it already exists. It so return the xref_id for this one.
##################################################################
- my $xref_id = $self->get_xref($acc,$source_id, $species_id);
+ my $xref_id = $self->get_xref($acc,$source_id, $species_id, $dbi);
if(defined $xref_id){
return $xref_id;
}
-
- #######################################################################
- # If the statement handle for the insertion of xrefs does not exist yet
- # then create it
- #######################################################################
- if (!(defined $add_xref_sth) ) {
- $add_xref_sth =
- $self->dbi->prepare( 'INSERT INTO xref '
+ my $add_xref_sth =
+ $dbi->prepare( 'INSERT INTO xref '
. '(accession,version,label,description,source_id,species_id, info_type, info_text) '
. 'VALUES(?,?,?,?,?,?,?,?)' );
- }
######################################################################
# If the description is more than 255 characters, chop it off and add
@@ -962,26 +948,23 @@ sub add_object_xref {
my $xref_id = $arg_ref->{xref_id} || croak 'add_object_xref needs an xref_id';
my $ensembl_id = $arg_ref->{ensembl_id} || croak 'add_object_xref needs a ensembl_id';
my $object_type = $arg_ref->{object_type} || croak 'add_object_xref needs an object_type';
+ my $dbi = $arg_ref->{dbi};
+
+ $dbi = $self->dbi unless defined $dbi;
##################################################################
# See if it already exists. It so return the xref_id for this one.
##################################################################
- my $object_xref_id = $self->get_object_xref($xref_id, $ensembl_id, $object_type);
+ my $object_xref_id = $self->get_object_xref($xref_id, $ensembl_id, $object_type, $dbi);
if(defined $object_xref_id){
return $object_xref_id;
}
- #######################################################################
- # If the statement handle for the insertion of object_xrefs does not exist yet
- # then create it
- #######################################################################
- if (!(defined $add_object_xref_sth) ) {
- $add_object_xref_sth =
- $self->dbi->prepare( 'INSERT INTO object_xref'
+ my $add_object_xref_sth =
+ $dbi->prepare( 'INSERT INTO object_xref'
. '(ensembl_id, ensembl_object_type, xref_id) '
. 'VALUES(?,?,?)' );
- }
####################################
# Add the object_xref and croak if it fails
@@ -1003,17 +986,14 @@ sub add_identity_xref {
my $score = $arg_ref->{score} || croak 'add_identity_xref needs a score';
my $target_identity = $arg_ref->{target_identity} || croak 'add_identity_xref needs a target_identity';
my $query_identity = $arg_ref->{query_identity} || croak 'add_identity_xref needs a query_identity';
+ my $dbi = $arg_ref->{dbi};
- #######################################################################
- # If the statement handle for the insertion of object_xrefs does not exist yet
- # then create it
- #######################################################################
- if (!(defined $add_identity_xref_sth) ) {
- $add_identity_xref_sth =
- $self->dbi->prepare( 'INSERT INTO identity_xref'
+ $dbi = $self->dbi unless defined $dbi;
+
+ my $add_identity_xref_sth =
+ $dbi->prepare( 'INSERT INTO identity_xref'
. '(object_xref_id, score, query_identity, target_identity) '
. 'VALUES(?,?,?,?)' );
- }
####################################
# Add the object_xref and croak if it fails
@@ -1041,24 +1021,20 @@ sub add_to_direct_xrefs{
my $description = $arg_ref->{desc};
my $linkage = $arg_ref->{linkage};
my $info_text = $arg_ref->{info_text} || '';
+ my $dbi = $arg_ref->{dbi};
- my $dbi= $self->dbi();
+ $dbi = $self->dbi unless defined $dbi;
- ######################
- # Get statement handle
- ######################
- if(!(defined $add_xref_sth)){
- my $sql = (<<'AXX');
+ my $sql = (<<'AXX');
INSERT INTO xref (accession,version,label,description,source_id,species_id, info_type, info_text)
VALUES (?,?,?,?,?,?,?,?)
AXX
- $add_xref_sth = $dbi->prepare($sql);
- }
+ my $add_xref_sth = $dbi->prepare($sql);
###############################################################
# If the acc already has an xrefs find it else cretae a new one
###############################################################
- my $direct_id = $self->get_xref($acc, $source_id, $species_id);
+ my $direct_id = $self->get_xref($acc, $source_id, $species_id, $dbi);
if(!(defined $direct_id)){
$add_xref_sth->execute(
$acc, $version || 0, $label,
@@ -1066,7 +1042,7 @@ AXX
) or croak("$acc\t$label\t\t$source_id\t$species_id\n");
}
- $direct_id = $self->get_xref($acc, $source_id, $species_id);
+ $direct_id = $self->get_xref($acc, $source_id, $species_id, $dbi);
#########################
# Now add the direct info
@@ -1084,33 +1060,12 @@ sub add_direct_xref {
my ($self, $general_xref_id, $ensembl_stable_id, $ensembl_type, $linkage_type, $dbi) = @_;
$dbi = $self->dbi unless defined $dbi;
- #######################################################
- # Create statement handles if they do not exist already
- ########################################################
- if (!(defined $add_direct_xref_sth{$ensembl_type})){
- my $add_gene_direct_xref_sth = $dbi->prepare('INSERT INTO gene_direct_xref VALUES(?,?,?)');
- my $add_tr_direct_xref_sth = $dbi->prepare('INSERT INTO transcript_direct_xref VALUES(?,?,?)');
- my $add_tl_direct_xref_sth = $dbi->prepare('INSERT INTO translation_direct_xref VALUES(?,?,?)');
- $add_direct_xref_sth{'gene'} = $add_gene_direct_xref_sth;
- $add_direct_xref_sth{'transcript'} = $add_tr_direct_xref_sth;
- $add_direct_xref_sth{'translation'} = $add_tl_direct_xref_sth;
- $add_direct_xref_sth{'Gene'} = $add_gene_direct_xref_sth;
- $add_direct_xref_sth{'Transcript'} = $add_tr_direct_xref_sth;
- $add_direct_xref_sth{'Translation'} = $add_tl_direct_xref_sth;
- }
- ##############################
- # Make sure type is recognised
- ##############################
- if(!(defined $add_direct_xref_sth{$ensembl_type})){
- croak "ERROR add_direct_xref_sth does not exist for $ensembl_type ???";
- }
- else{
- ##########################
- # Add the direct xref data
- ##########################
- $add_direct_xref_sth{$ensembl_type}->execute($general_xref_id, $ensembl_stable_id, $linkage_type);
- }
+ $ensembl_type = lc($ensembl_type);
+ my $sql = "INSERT INTO " . $ensembl_type . "_direct_xref VALUES (?,?,?)";
+ my $add_direct_xref_sth = $dbi->prepare($sql);
+
+ $add_direct_xref_sth->execute($general_xref_id, $ensembl_stable_id, $linkage_type);
return;
}
@@ -1130,41 +1085,35 @@ sub add_dependent_xref{
my $description = $arg_ref->{desc};
my $linkage = $arg_ref->{linkage};
my $info_text = $arg_ref->{info_text} || '';
+ my $dbi = $arg_ref->{dbi};
- my $dbi = $self->dbi;
+ $dbi = $self->dbi unless defined $dbi;
- ########################################
- # Create/Get the statement handle needed
- ########################################
- if(!(defined $add_xref_sth)){
- my $sql = (<<'IXR');
+ my $sql = (<<'IXR');
INSERT INTO xref
(accession,version,label,description,source_id,species_id, info_type, info_text)
VALUES (?,?,?,?,?,?,?,?)
IXR
- $add_xref_sth = $dbi->prepare($sql);
- }
- if(!(defined $add_dependent_xref_sth)){
- my $sql = (<<'ADX');
+ my $add_xref_sth = $dbi->prepare($sql);
+ $sql = (<<'ADX');
INSERT INTO dependent_xref
(master_xref_id,dependent_xref_id,linkage_annotation,linkage_source_id)
VALUES (?,?,?,?)
ADX
- $add_dependent_xref_sth = $dbi->prepare($sql);
- }
+ my $add_dependent_xref_sth = $dbi->prepare($sql);
####################################################
# Does the xref already exist. If so get its xref_id
# else create it and get the new xref_id
####################################################
- my $dependent_id = $self->get_xref($acc, $source_id, $species_id);
+ my $dependent_id = $self->get_xref($acc, $source_id, $species_id, $dbi);
if(!(defined $dependent_id)){
$add_xref_sth->execute(
$acc, $version, $label,
$description, $source_id, $species_id, 'DEPENDENT', $info_text
) or croak("$acc\t$label\t\t$source_id\t$species_id\n");
}
- $dependent_id = $self->get_xref($acc, $source_id, $species_id);
+ $dependent_id = $self->get_xref($acc, $source_id, $species_id, $dbi);
################################################
# Croak if we have failed to create.get the xref
@@ -1193,15 +1142,13 @@ ADX
# but want to write synonyms for each with the same accession
##################################################################
sub add_to_syn_for_mult_sources{
- my ($self, $acc, $sources, $syn, $species_id) = @_;
+ my ($self, $acc, $sources, $syn, $species_id, $dbi) = @_;
- my $dbi = $self->dbi;
- if(!(defined $add_synonym_sth)){
- $add_synonym_sth = $dbi->prepare('INSERT IGNORE INTO synonym VALUES(?,?)');
- }
+ $dbi = $self->dbi unless defined $dbi;
+ my $add_synonym_sth = $dbi->prepare('INSERT IGNORE INTO synonym VALUES(?,?)');
foreach my $source_id (@{$sources}){
- my $xref_id = $self->get_xref($acc, $source_id, $species_id);
+ my $xref_id = $self->get_xref($acc, $source_id, $species_id, $dbi);
if(defined $xref_id){
$add_synonym_sth->execute( $xref_id, $syn )
or croak( $dbi->errstr() . "\n $xref_id\n $syn\n" );
@@ -1215,13 +1162,11 @@ sub add_to_syn_for_mult_sources{
# Add synomyn for an xref given by accession and source_id
##########################################################
sub add_to_syn{
- my ($self, $acc, $source_id, $syn, $species_id) = @_;
+ my ($self, $acc, $source_id, $syn, $species_id, $dbi) = @_;
- my $dbi = $self->dbi;
- if(!(defined $add_synonym_sth)){
- $add_synonym_sth = $dbi->prepare('INSERT IGNORE INTO synonym VALUES(?,?)');
- }
- my $xref_id = $self->get_xref($acc, $source_id, $species_id);
+ $dbi = $self->dbi unless defined $dbi;
+ my $add_synonym_sth = $dbi->prepare('INSERT IGNORE INTO synonym VALUES(?,?)');
+ my $xref_id = $self->get_xref($acc, $source_id, $species_id, $dbi);
if(defined $xref_id){
$add_synonym_sth->execute( $xref_id, $syn )
or croak( $dbi->errstr() . "\n $xref_id\n $syn\n" );
@@ -1238,12 +1183,10 @@ sub add_to_syn{
# Add synomyn for an xref given by xref_id
##########################################
sub add_synonym{
- my ($self, $xref_id, $syn) = @_;
+ my ($self, $xref_id, $syn, $dbi) = @_;
- my $dbi=$self->dbi;
- if(!(defined $add_synonym_sth)){
- $add_synonym_sth = $dbi->prepare('INSERT IGNORE INTO synonym VALUES(?,?)');
- }
+ $dbi = $self->dbi unless defined $dbi;
+ my $add_synonym_sth = $dbi->prepare('INSERT IGNORE INTO synonym VALUES(?,?)');
$add_synonym_sth->execute( $xref_id, $syn )
or croak( $dbi->errstr()."\n $xref_id\n $syn\n\n" );
@@ -1256,9 +1199,11 @@ sub add_synonym{
# as keys.
#######################################################
sub get_label_to_acc{
- my ($self, $name, $species_id, $prio_desc) = @_;
+ my ($self, $name, $species_id, $prio_desc, $dbi) = @_;
my %hash1=();
+ $dbi = $self->dbi unless defined $dbi;
+
my $sql =(<<"GLA");
SELECT xref.accession, xref.label
FROM xref, source
@@ -1271,7 +1216,7 @@ GLA
if(defined $species_id){
$sql .= " and xref.species_id = $species_id";
}
- my $sub_sth = $self->dbi->prepare($sql);
+ my $sub_sth = $dbi->prepare($sql);
$sub_sth->execute();
while(my @row = $sub_sth->fetchrow_array()) {
@@ -1297,7 +1242,7 @@ GLS
if(defined $species_id){
$sql .= " AND xref.species_id = $species_id";
}
- $sub_sth = $self->dbi->prepare($sql);
+ $sub_sth = $dbi->prepare($sql);
$sub_sth->execute();
while(my @row = $sub_sth->fetchrow_array()) {
@@ -1312,9 +1257,11 @@ GLS
# and the label as the value.
#######################################################
sub get_acc_to_label{
- my ($self, $name, $species_id, $prio_desc) = @_;
+ my ($self, $name, $species_id, $prio_desc, $dbi) = @_;
my %hash1=();
+ $dbi = $self->dbi unless defined $dbi;
+
my $sql =(<<"GLA");
SELECT xref.accession, xref.label
FROM xref, source
@@ -1327,7 +1274,7 @@ GLA
if(defined $species_id){
$sql .= " and xref.species_id = $species_id";
}
- my $sub_sth = $self->dbi->prepare($sql);
+ my $sub_sth = $dbi->prepare($sql);
$sub_sth->execute();
while(my @row = $sub_sth->fetchrow_array()) {
@@ -1344,10 +1291,10 @@ GLA
# as keys.
#######################################################
sub get_label_to_desc{
- my ($self, $name, $species_id, $prio_desc) = @_;
+ my ($self, $name, $species_id, $prio_desc, $dbi) = @_;
my %hash1=();
- my $dbi = $self->dbi();
+ $dbi = $self->dbi unless defined $dbi;
my $sql =(<<"GDH");
SELECT xref.description, xref.label
@@ -1401,10 +1348,12 @@ GDS
# Set release for a particular source_id.
########################################
sub set_release{
- my ($self, $source_id, $s_release ) = @_;
+ my ($self, $source_id, $s_release, $dbi ) = @_;
+
+ $dbi = $self->dbi unless defined $dbi;
my $sth =
- $self->dbi->prepare('UPDATE source SET source_release=? WHERE source_id=?');
+ $dbi->prepare('UPDATE source SET source_release=? WHERE source_id=?');
if($verbose) { print "Setting release to '$s_release' for source ID '$source_id'\n"; }
@@ -1420,6 +1369,9 @@ sub set_release{
sub get_dependent_mappings {
my $self = shift;
my $source_id = shift;
+ my $dbi = shift;
+
+ $dbi = $self->dbi unless defined $dbi;
my $sql =(<<"GDM");
SELECT d.master_xref_id, d.dependent_xref_id
@@ -1427,7 +1379,7 @@ sub get_dependent_mappings {
WHERE x.xref_id = d.dependent_xref_id AND
x.source_id = $source_id
GDM
- my $sth = $self->dbi->prepare($sql);
+ my $sth = $dbi->prepare($sql);
$sth->execute();
my $master_xref;
my $dependent_xref;
@@ -1447,6 +1399,8 @@ GDM
sub get_ext_synonyms{
my $self = shift;
my $source_name = shift;
+ my $dbi = shift;
+ $dbi = $self->dbi unless defined $dbi;
my %ext_syns;
my %seen; # can be in more than once fro each type of external source.
my $separator = qw{:};
@@ -1458,7 +1412,7 @@ sub get_ext_synonyms{
so.source_id = x.source_id AND
so.name like '$source_name'
GES
- my $sth = $self->dbi->prepare($sql);
+ my $sth = $dbi->prepare($sql);
$sth->execute;
my ($acc, $label, $syn);
@@ -1485,6 +1439,8 @@ GES
######################################################################
sub parsing_finished_store_data {
my $self = shift;
+ my $dbi = shift;
+ $dbi = $self->dbi unless defined $dbi;
# Store max id for
@@ -1500,7 +1456,7 @@ sub parsing_finished_store_data {
( 'xref' => 'xref_id', 'object_xref' => 'object_xref_id' );
foreach my $table ( keys %table_and_key ) {
- my $sth = $self->dbi->prepare(
+ my $sth = $dbi->prepare(
'select MAX(' . $table_and_key{$table} . ") from $table" );
$sth->execute;
my $max_val;
@@ -1508,16 +1464,17 @@ sub parsing_finished_store_data {
$sth->fetch;
$sth->finish;
$self->add_meta_pair( 'PARSED_' . $table_and_key{$table},
- $max_val || 1 );
+ $max_val || 1, $dbi );
}
return;
} ## end sub parsing_finished_store_data
sub get_meta_value {
- my ($self, $key) = @_;
+ my ($self, $key, $dbi) = @_;
+ $dbi = $self->dbi unless defined $dbi;
- my $sth = $self->dbi->prepare('select meta_value from meta where meta_key like "'.$key.'" order by meta_id');
+ my $sth = $dbi->prepare('select meta_value from meta where meta_key like "'.$key.'" order by meta_id');
$sth->execute();
my $value;
$sth->bind_columns(\$value);
diff --git a/misc-scripts/xref_mapping/XrefParser/CCDSParser.pm b/misc-scripts/xref_mapping/XrefParser/CCDSParser.pm
index 26710ddcb6..be3f0e3c62 100644
--- a/misc-scripts/xref_mapping/XrefParser/CCDSParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/CCDSParser.pm
@@ -37,6 +37,8 @@ sub run_script {
my $file = $ref_arg->{file};
my $verbose = $ref_arg->{verbose};
my $db = $ref_arg->{dba};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $file) ){
croak "Need to pass source_id, species_id and file as pairs";
@@ -106,6 +108,7 @@ SCD
label => $display_label,
source_id => $source_id,
species_id => $species_id,
+ dbi => $dbi,
info_type => "DIRECT"} );
$xref_count++;
$seen{$display_label} = $xref_id;
@@ -114,7 +117,7 @@ SCD
$xref_id = $seen{$display_label};
}
- $self->add_direct_xref($xref_id, $stable_id, "Transcript", "");
+ $self->add_direct_xref($xref_id, $stable_id, "Transcript", "", $dbi);
$direct_count++;
}
diff --git a/misc-scripts/xref_mapping/XrefParser/CoreXrefParser.pm b/misc-scripts/xref_mapping/XrefParser/CoreXrefParser.pm
index 254f3d1a78..46113b19cc 100644
--- a/misc-scripts/xref_mapping/XrefParser/CoreXrefParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/CoreXrefParser.pm
@@ -34,6 +34,8 @@ sub run_script {
my $species_id = $ref_arg->{species_id};
my $file = $ref_arg->{file};
my $verbose = $ref_arg->{verbose};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $file) ){
croak "Need to pass source_id, species_id and file as pairs";
@@ -62,7 +64,7 @@ sub run_script {
$copy_description_from_object = $1;
}
- my $external_db_name = $self->get_source_name_for_source_id($source_id);
+ my $external_db_name = $self->get_source_name_for_source_id($source_id, $dbi);
#copy object xrefs from core
@@ -96,7 +98,7 @@ sub run_script {
}
#get the species name
- my %id2name = $self->species_id2name;
+ my %id2name = $self->species_id2name($dbi);
my $species_name = $id2name{$species_id}[0];
if (!$object_type) {
@@ -173,7 +175,7 @@ sub run_script {
# but we need to attach them specifcally
# to RNAmmer, tRNAScan or RFAM
# so get the source based on the db_name from the core db
- my $external_source_id = $self->get_source_id_for_source_name($db_name);
+ my $external_source_id = $self->get_source_id_for_source_name($db_name, undef, $dbi);
if (! defined $external_source_id) {
warn ("can't get a source_id for external_db, $db_name!\n");
@@ -200,6 +202,7 @@ sub run_script {
desc => $description,
source_id => $external_source_id,
species_id => $species_id,
+ dbi => $dbi,
info_type => "DIRECT"} );
@@ -210,7 +213,7 @@ sub run_script {
$xref_id = $added_xref{$xref->primary_id()};
}
- $self->add_direct_xref($xref_id, $object->stable_id(), $valid_object_types{$object_type}, "");
+ $self->add_direct_xref($xref_id, $object->stable_id(), $valid_object_types{$object_type}, "", $dbi);
$direct_count++;
}
}
diff --git a/misc-scripts/xref_mapping/XrefParser/DBASSParser.pm b/misc-scripts/xref_mapping/XrefParser/DBASSParser.pm
index b81211b0ec..c229ce9b05 100644
--- a/misc-scripts/xref_mapping/XrefParser/DBASSParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/DBASSParser.pm
@@ -44,6 +44,8 @@ sub run {
my $species_id = $ref_arg->{species_id};
my $files = $ref_arg->{files};
my $verbose = $ref_arg->{verbose};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $files) ){
croak "Need to pass source_id, species_id and files as pairs";
@@ -101,7 +103,7 @@ sub run {
++$parsed_count;
- my $xref_id = $self->get_xref( $dbass_gene_id, $source_id, $species_id );
+ my $xref_id = $self->get_xref( $dbass_gene_id, $source_id, $species_id, $dbi );
if ( !defined($xref_id) || $xref_id eq '' ) {
$xref_id = $self->add_xref({ acc => $dbass_gene_id,
@@ -109,14 +111,15 @@ sub run {
label => $label,
desc => $description,
source_id => $source_id,
+ dbi => $dbi,
species_id => $species_id,
info_type => "DIRECT"} );
}
- $self->add_direct_xref( $xref_id, $ensembl_id, $type, '');
+ $self->add_direct_xref( $xref_id, $ensembl_id, $type, '', $dbi);
if (defined ($synonym)) {
- $self->add_synonym($xref_id, $synonym);
+ $self->add_synonym($xref_id, $synonym, $dbi);
}
elsif ($synonym =~ /^\s/){
print "There is white space! \n" if($verbose);
diff --git a/misc-scripts/xref_mapping/XrefParser/EntrezGeneParser.pm b/misc-scripts/xref_mapping/XrefParser/EntrezGeneParser.pm
index 9a6fef0082..5a4ba1e9b2 100644
--- a/misc-scripts/xref_mapping/XrefParser/EntrezGeneParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/EntrezGeneParser.pm
@@ -32,6 +32,8 @@ sub run {
my $species_id = $ref_arg->{species_id};
my $files = $ref_arg->{files};
my $verbose = $ref_arg->{verbose};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $files) ){
croak "Need to pass source_id, species_id, files and rel_file as pairs";
@@ -40,9 +42,9 @@ sub run {
my $file = @{$files}[0];
- my $wiki_source_id = $self->get_source_id_for_source_name("WikiGene");
+ my $wiki_source_id = $self->get_source_id_for_source_name("WikiGene", undef, $dbi);
- my %species_tax_id = %{$self->get_taxonomy_from_species_id($species_id)};
+ my %species_tax_id = %{$self->get_taxonomy_from_species_id($species_id, $dbi)};
my $eg_io = $self->get_filehandle($file);
@@ -120,6 +122,7 @@ sub run {
desc => $desc,
source_id => $source_id,
species_id => $species_id,
+ dbi => $dbi,
info_type =>"DEPENDENT"} );
$self->add_xref({ acc => $acc,
@@ -127,13 +130,14 @@ sub run {
desc => $desc,
source_id => $wiki_source_id,
species_id => $species_id,
+ dbi => $dbi,
info_type => "DEPENDENT" } ); #,"From EntrezGene $acc");
$xref_count++;
my (@syn) = split(/\|/ ,$arr[$gene_synonyms_index]);
foreach my $synonym (@syn){
if($synonym ne "-"){
- $self->add_to_syn($acc, $source_id, $synonym, $species_id);
+ $self->add_to_syn($acc, $source_id, $synonym, $species_id, $dbi);
$syn_count++;
}
}
diff --git a/misc-scripts/xref_mapping/XrefParser/HGNCParser.pm b/misc-scripts/xref_mapping/XrefParser/HGNCParser.pm
index 814a7c578b..3ec3d6be67 100644
--- a/misc-scripts/xref_mapping/XrefParser/HGNCParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/HGNCParser.pm
@@ -33,6 +33,8 @@ sub run_script {
my $file = $ref_arg->{file};
my $verbose = $ref_arg->{verbose};
my $db = $ref_arg->{dba};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $file) ){
croak "Need to pass source_id, species_id, file as pairs";
@@ -87,14 +89,14 @@ sub run_script {
}
- my (%swissprot) = %{$self->get_valid_codes('Uniprot/SWISSPROT',$species_id)};
- my (%refseq) = %{$self->get_valid_codes('refseq',$species_id)};
+ my (%swissprot) = %{$self->get_valid_codes('Uniprot/SWISSPROT',$species_id, $dbi)};
+ my (%refseq) = %{$self->get_valid_codes('refseq',$species_id, $dbi)};
my @list;
push @list, 'refseq_peptide';
push @list, 'refseq_mRNA';
- my (%entrezgene) = %{$self->get_valid_xrefs_for_dependencies('EntrezGene',@list)};
- my $source_name = $self->get_source_name_for_source_id($source_id);
- my $name_to_source_id = $self->get_sources($source_name);
+ my (%entrezgene) = %{$self->get_valid_xrefs_for_dependencies('EntrezGene', $dbi, @list)};
+ my $source_name = $self->get_source_name_for_source_id($source_id, $dbi);
+ my $name_to_source_id = $self->get_sources($source_name, $dbi);
my %name_count;
@@ -157,11 +159,13 @@ sub run_script {
label => $symbol,
desc => $name,
source_id => $source_id,
+ dbi => $dbi,
species_id => $species_id} );
$self->add_synonyms_for_hgnc( {source_id => $source_id,
name => $acc,
species_id => $species_id,
+ dbi => $dbi,
dead => $previous_symbols,
alias => $synonyms} );
$name_count{$type}++;
@@ -187,11 +191,13 @@ sub run_script {
label => $symbol,
desc => $name,
source_id => $source_id,
+ dbi => $dbi,
species_id => $species_id} );
$self->add_synonyms_for_hgnc( {source_id => $source_id,
name => $acc,
species_id => $species_id,
+ dbi => $dbi,
dead => $previous_symbols,
alias => $synonyms} );
$name_count{$type}++;
@@ -211,7 +217,8 @@ sub run_script {
type => 'gene',
acc => $acc,
label => $symbol,
- desc => $name,,
+ desc => $name,
+ dbi => $dbi,
source_id => $source_id,
species_id => $species_id} );
@@ -219,6 +226,7 @@ sub run_script {
name => $acc,
species_id => $species_id,
dead => $previous_symbols,
+ dbi => $dbi,
alias => $synonyms});
}
@@ -236,11 +244,13 @@ sub run_script {
label => $symbol,
desc => $name || '',
source_id => $source_id,
+ dbi => $dbi,
species_id => $species_id} );
}
$self->add_synonyms_for_hgnc( {source_id => $source_id,
name => $acc,
species_id => $species_id,
+ dbi => $dbi,
dead => $previous_symbols,
alias => $synonyms});
}
@@ -260,12 +270,14 @@ sub run_script {
label => $symbol,
desc => $name || '',
source_id => $source_id,
+ dbi => $dbi,
species_id => $species_id} );
$name_count{$type}++;
$self->add_synonyms_for_hgnc( {source_id => $source_id,
name => $acc,
species_id => $species_id,
dead => $previous_symbols,
+ dbi => $dbi,
alias => $synonyms});
}
}
@@ -278,11 +290,13 @@ sub run_script {
desc => $name,
source_id => $source_id,
species_id => $species_id,
+ dbi => $dbi,
info_type => "MISC"} );
$self->add_synonyms_for_hgnc( {source_id => $source_id,
name => $acc,
species_id => $species_id,
+ dbi => $dbi,
dead => $previous_symbols,
alias => $synonyms});
$mismatch++;
@@ -305,20 +319,21 @@ sub run_script {
sub get_sources {
my $self = shift;
my $source_name = shift;
+ my $dbi = shift;
my %name_to_source_id;
my @sources = ('entrezgene_manual', 'refseq_manual', 'entrezgene_mapped', 'refseq_mapped', 'ensembl_manual', 'swissprot_manual', 'desc_only', 'ccds');
foreach my $key (@sources) {
- my $source_id = $self->get_source_id_for_source_name($source_name, $key);
+ my $source_id = $self->get_source_id_for_source_name($source_name, $key, $dbi);
if(!(defined $source_id)){
die 'Could not get source id for HGNC and '. $key ."\n";
}
$name_to_source_id{ $key } = $source_id;
}
- my $source_id = $self->get_source_id_for_source_name('LRG_HGNC_notransfer');
+ my $source_id = $self->get_source_id_for_source_name('LRG_HGNC_notransfer', undef, $dbi);
if(!(defined $source_id) ){
die 'Could not get source id for LRG_HGNC_notransfer\n';
}
@@ -335,18 +350,19 @@ sub add_synonyms_for_hgnc{
my $species_id = $ref_arg->{species_id};
my $dead_name = $ref_arg->{dead};
my $alias = $ref_arg->{alias};
+ my $dbi = $ref_arg->{dbi};
if (defined $dead_name ) { # dead name, add to synonym
my @array2 = split ',\s*', $dead_name ;
foreach my $arr (@array2){
- $self->add_to_syn($name, $source_id, $arr, $species_id);
+ $self->add_to_syn($name, $source_id, $arr, $species_id, $dbi);
}
}
if (defined $alias ) { # alias, add to synonym
my @array2 = split ',\s*', $alias;
foreach my $arr (@array2){
- $self->add_to_syn($name, $source_id, $arr, $species_id);
+ $self->add_to_syn($name, $source_id, $arr, $species_id, $dbi);
}
}
return;
diff --git a/misc-scripts/xref_mapping/XrefParser/HGNC_CCDSParser.pm b/misc-scripts/xref_mapping/XrefParser/HGNC_CCDSParser.pm
index 40f7e76b8f..f6ea32982a 100644
--- a/misc-scripts/xref_mapping/XrefParser/HGNC_CCDSParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/HGNC_CCDSParser.pm
@@ -37,6 +37,8 @@ sub run_script {
my $species_id = $ref_arg->{species_id};
my $file = $ref_arg->{file};
my $verbose = $ref_arg->{verbose};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $file) ){
croak "Need to pass source_id, species_id and file as pairs";
@@ -148,12 +150,10 @@ SQL
my %version;
my %description;
- my $dbi = $self->dbi();
-
my $sql_syn = "insert ignore into synonym (xref_id, synonym) values (?, ?)";
my $add_syn_sth = $dbi->prepare($sql_syn);
- my $syn_hash = $self->get_ext_synonyms("HGNC");
+ my $syn_hash = $self->get_ext_synonyms("HGNC", $dbi);
$sql = 'select source_id, priority_description from source where name like "HGNC"';
$sth = $dbi->prepare($sql);
@@ -194,10 +194,11 @@ SQL
desc => $description{$hgnc},
source_id => $source_id,
species_id => $species_id,
+ dbi => $dbi,
info_type => "DIRECT"} );
foreach my $stable_id (@{$ccds_to_stable_id{$ccds}}){
- $self->add_direct_xref($xref_id, $stable_id, "Transcript", "");
+ $self->add_direct_xref($xref_id, $stable_id, "Transcript", "", $dbi);
$direct_count++;
}
diff --git a/misc-scripts/xref_mapping/XrefParser/HPAParser.pm b/misc-scripts/xref_mapping/XrefParser/HPAParser.pm
index 2d475a2ca0..7de5e6145b 100644
--- a/misc-scripts/xref_mapping/XrefParser/HPAParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/HPAParser.pm
@@ -40,6 +40,8 @@ sub run {
my $species_id = $ref_arg->{species_id};
my $files = $ref_arg->{files};
my $verbose = $ref_arg->{verbose};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $files) ){
croak "Need to pass source_id, species_id, files and rel_file as pairs";
@@ -78,7 +80,7 @@ sub run {
++$parsed_count;
- my $xref_id = $self->get_xref( $antibody_id, $source_id, $species_id );
+ my $xref_id = $self->get_xref( $antibody_id, $source_id, $species_id, $dbi );
if ( !defined($xref_id) || $xref_id eq '' ) {
$xref_id = $self->add_xref({ acc => $antibody_id,
@@ -87,11 +89,12 @@ sub run {
desc => $description,
source_id => $source_id,
species_id => $species_id,
+ dbi => $dbi,
info_type => "DIRECT"} );
}
- $self->add_direct_xref( $xref_id, $ensembl_peptide_id, $type, '');
+ $self->add_direct_xref( $xref_id, $ensembl_peptide_id, $type, '', $dbi);
} ## end while ( defined( my $line...
diff --git a/misc-scripts/xref_mapping/XrefParser/JGI_Parser.pm b/misc-scripts/xref_mapping/XrefParser/JGI_Parser.pm
index f4d71ee3c9..64d3896472 100644
--- a/misc-scripts/xref_mapping/XrefParser/JGI_Parser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/JGI_Parser.pm
@@ -40,6 +40,8 @@ sub run {
my $files = $ref_arg->{files};
my $release_file = $ref_arg->{rel_file};
my $verbose = $ref_arg->{verbose};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $files) ){
croak "Need to pass source_id, species_id and files as pairs";
@@ -49,7 +51,7 @@ sub run {
my $file = @{$files}[0];
- my $source_name = $self->get_source_name_for_source_id ($source_id) ;
+ my $source_name = $self->get_source_name_for_source_id ($source_id, $dbi) ;
# the source name defines how to parse the header
# different formats for different sources (all have entries in external_db.txt and populate_metadata.sql )
@@ -159,7 +161,7 @@ sub run {
print scalar(@xrefs) . " JGI_ xrefs succesfully parsed\n" if($verbose);
- $self->upload_xref_object_graphs(\@xrefs);
+ $self->upload_xref_object_graphs(\@xrefs, $dbi);
return 0; # successful
}
diff --git a/misc-scripts/xref_mapping/XrefParser/MGIParser.pm b/misc-scripts/xref_mapping/XrefParser/MGIParser.pm
index d80b67199b..49a13157ed 100644
--- a/misc-scripts/xref_mapping/XrefParser/MGIParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/MGIParser.pm
@@ -33,6 +33,8 @@ sub run {
my $species_id = $ref_arg->{species_id};
my $files = $ref_arg->{files};
my $verbose = $ref_arg->{verbose};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $files) ){
croak "Need to pass source_id, species_id and files as pairs";
@@ -51,9 +53,6 @@ sub run {
my %description;
my %accession;
-
- my $dbi = $self->dbi();
-
my $sql = 'select source_id, priority_description from source where name like "MGI"';
my $sth = $dbi->prepare($sql);
@@ -86,7 +85,7 @@ sub run {
$sql = "insert ignore into synonym (xref_id, synonym) values (?, ?)";
my $add_syn_sth = $dbi->prepare($sql);
- my $syn_hash = $self->get_ext_synonyms("MGI");
+ my $syn_hash = $self->get_ext_synonyms("MGI", $dbi);
my $count = 0;
my $syn_count = 0;
@@ -101,10 +100,11 @@ sub run {
label => $label{$acc},
desc => $description{$acc},
source_id => $source_id,
+ dbi => $dbi,
species_id => $species_id,
info_type => "DIRECT"} );
- $self->add_direct_xref( $xref_id, $ensid, "Gene", '');
+ $self->add_direct_xref( $xref_id, $ensid, "Gene", '', $dbi);
if(defined($syn_hash->{$acc})){
foreach my $syn (@{$syn_hash->{$acc}}){
$add_syn_sth->execute($xref_id, $syn);
diff --git a/misc-scripts/xref_mapping/XrefParser/MGI_CCDS_Parser.pm b/misc-scripts/xref_mapping/XrefParser/MGI_CCDS_Parser.pm
index ce7beb46fe..85400a547c 100644
--- a/misc-scripts/xref_mapping/XrefParser/MGI_CCDS_Parser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/MGI_CCDS_Parser.pm
@@ -34,6 +34,8 @@ sub run_script {
my $species_id = $ref_arg->{species_id};
my $file = $ref_arg->{file};
my $verbose = $ref_arg->{verbose};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $file) ){
croak "Need to pass source_id, species_id and file as pairs";
@@ -52,8 +54,6 @@ sub run_script {
my %description;
my %accession;
- my $dbi = $self->dbi();
-
my $sql = 'select source_id, priority_description from source where name like "MGI"';
my $sth = $dbi->prepare($sql);
@@ -135,6 +135,7 @@ sub run_script {
label => $label{$acc},
desc => $description{$acc},
source_id => $source_id,
+ dbi => $dbi,
species_id => $species_id });
$count++;
diff --git a/misc-scripts/xref_mapping/XrefParser/MGI_Desc_Parser.pm b/misc-scripts/xref_mapping/XrefParser/MGI_Desc_Parser.pm
index 44822308ae..6636fab174 100644
--- a/misc-scripts/xref_mapping/XrefParser/MGI_Desc_Parser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/MGI_Desc_Parser.pm
@@ -36,6 +36,8 @@ sub run {
my $species_id = $ref_arg->{species_id};
my $files = $ref_arg->{files};
my $verbose = $ref_arg->{verbose};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $files) ){
croak "Need to pass source_id, species_id and files as pairs";
@@ -84,6 +86,7 @@ sub run {
desc => $desc,
source_id => $source_id,
species_id => $species_id,
+ dbi => $dbi,
info_type => "MISC"} );
if($verbose and !$desc){
print "$accession has no description\n";
@@ -93,7 +96,7 @@ sub run {
if(defined($acc_to_xref{$accession})){
foreach my $syn (@synonyms) {
- $self->add_synonym($acc_to_xref{$accession}, $syn);
+ $self->add_synonym($acc_to_xref{$accession}, $syn, $dbi);
$syn_count++;
}
diff --git a/misc-scripts/xref_mapping/XrefParser/MIMParser.pm b/misc-scripts/xref_mapping/XrefParser/MIMParser.pm
index 4f0f70b86a..82d637308f 100644
--- a/misc-scripts/xref_mapping/XrefParser/MIMParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/MIMParser.pm
@@ -34,6 +34,8 @@ sub run {
my $species_id = $ref_arg->{species_id};
my $files = $ref_arg->{files};
my $verbose = $ref_arg->{verbose};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $general_source_id) or (!defined $species_id) or (!defined $files) ){
croak "Need to pass source_id, species_id and files as pairs";
@@ -50,9 +52,9 @@ sub run {
push @sources, $general_source_id;
- my $gene_source_id = $self->get_source_id_for_source_name("MIM_GENE");
+ my $gene_source_id = $self->get_source_id_for_source_name("MIM_GENE", undef, $dbi);
push @sources, $gene_source_id;
- my $morbid_source_id = $self->get_source_id_for_source_name("MIM_MORBID");
+ my $morbid_source_id = $self->get_source_id_for_source_name("MIM_MORBID", undef, $dbi);
push @sources, $morbid_source_id;
print "sources are:- ".join(", ",@sources)."\n" if($verbose);
@@ -111,6 +113,7 @@ sub run {
desc => $long_desc,
source_id => $gene_source_id,
species_id => $species_id,
+ dbi => $dbi,
info_type => "DEPENDENT"} );
}
elsif((!defined $type) or ($type eq "") or ($type eq "#") or ($type eq "%")){ #phenotype only
@@ -120,6 +123,7 @@ sub run {
desc => $long_desc,
source_id => $morbid_source_id,
species_id => $species_id,
+ dbi => $dbi,
info_type => "DEPENDENT"} );
}
elsif($type eq "+"){ # both
@@ -130,6 +134,7 @@ sub run {
desc => $long_desc,
source_id => $gene_source_id,
species_id => $species_id,
+ dbi => $dbi,
info_type => "DEPENDENT"} );
$self->add_xref({ acc => $number,
@@ -137,6 +142,7 @@ sub run {
desc => $long_desc,
source_id => $morbid_source_id,
species_id => $species_id,
+ dbi => $dbi,
info_type => "DEPENDENT"} );
}
elsif($type eq "^"){
@@ -164,7 +170,7 @@ sub run {
$new = $old_to_new{$new};
}
if(!defined($removed{$new})){
- $self->add_to_syn_for_mult_sources($new, \@sources, $old, $species_id);
+ $self->add_to_syn_for_mult_sources($new, \@sources, $old, $species_id, $dbi);
$syn_count++;
}
}
diff --git a/misc-scripts/xref_mapping/XrefParser/Mim2GeneParser.pm b/misc-scripts/xref_mapping/XrefParser/Mim2GeneParser.pm
index 13991cf807..5292cb972a 100644
--- a/misc-scripts/xref_mapping/XrefParser/Mim2GeneParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/Mim2GeneParser.pm
@@ -34,6 +34,8 @@ sub run {
my $species_id = $ref_arg->{species_id};
my $files = $ref_arg->{files};
my $verbose = $ref_arg->{verbose};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $general_source_id) or (!defined $species_id) or (!defined $files) ){
croak "Need to pass source_id, species_id and files as pairs";
@@ -51,14 +53,13 @@ sub run {
my $entrez_source_id =
$self->get_source_id_for_source_name(
- 'EntrezGene');
+ 'EntrezGene', undef, $dbi);
- my (%mim_gene) = %{$self->get_valid_codes("MIM_GENE",$species_id)};
- my (%mim_morbid) = %{$self->get_valid_codes("MIM_MORBID",$species_id)};
- my (%entrez) = %{$self->get_valid_codes("EntrezGene",$species_id)};
+ my (%mim_gene) = %{$self->get_valid_codes("MIM_GENE",$species_id, $dbi)};
+ my (%mim_morbid) = %{$self->get_valid_codes("MIM_MORBID",$species_id, $dbi)};
+ my (%entrez) = %{$self->get_valid_codes("EntrezGene",$species_id, $dbi)};
- my $dbi = $self->dbi();
- my $add_dependent_xref_sth = $self->dbi->prepare("INSERT INTO dependent_xref (master_xref_id,dependent_xref_id, linkage_source_id) VALUES (?,?, $entrez_source_id)");
+ my $add_dependent_xref_sth = $dbi->prepare("INSERT INTO dependent_xref (master_xref_id,dependent_xref_id, linkage_source_id) VALUES (?,?, $entrez_source_id)");
my $missed_entrez = 0;
my $missed_omim = 0;
diff --git a/misc-scripts/xref_mapping/XrefParser/PomBaseParser.pm b/misc-scripts/xref_mapping/XrefParser/PomBaseParser.pm
index f28d89f18c..475bfd225d 100644
--- a/misc-scripts/xref_mapping/XrefParser/PomBaseParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/PomBaseParser.pm
@@ -37,6 +37,8 @@ sub run {
my $species_id = $ref_arg->{species_id};
my $files = $ref_arg->{files};
my $verbose = $ref_arg->{verbose};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $files) ){
croak "Need to pass source_id, species_id and files as pairs";
@@ -45,8 +47,8 @@ sub run {
my $file = @{$files}[0];
- my $gene_source_id = $self->get_source_id_for_source_name("PomBase_GENE");
- my $transcript_source_id = $self->get_source_id_for_source_name("PomBase_TRANSCRIPT");
+ my $gene_source_id = $self->get_source_id_for_source_name("PomBase_GENE", undef, $dbi);
+ my $transcript_source_id = $self->get_source_id_for_source_name("PomBase_TRANSCRIPT", undef, $dbi);
my $pombase_io = $self->get_filehandle($file);
@@ -88,18 +90,20 @@ sub run {
desc => $desc,
source_id => $gene_source_id,
species_id => $species_id,
+ dbi => $dbi,
info_type => $info_type} );
- $self->add_direct_xref($ensembl_xref_id, $pombase_id, $ensembl_object_type, $info_type);
+ $self->add_direct_xref($ensembl_xref_id, $pombase_id, $ensembl_object_type, $info_type, $dbi);
} elsif ($ensembl_object_type eq 'Transcript') {
my $ensembl_xref_id = $self->add_xref({ acc => $pombase_id,
label => $name,
desc => $desc,
+ dbi => $dbi,
source_id => $transcript_source_id,
species_id => $species_id,
info_type => $info_type} );
- $self->add_direct_xref($ensembl_xref_id, $pombase_id, $ensembl_object_type, $info_type);
+ $self->add_direct_xref($ensembl_xref_id, $pombase_id, $ensembl_object_type, $info_type, $dbi);
}
$xref_count++;
@@ -109,7 +113,7 @@ sub run {
if ($verbose) {
print STDERR "adding synonym, $synonym\n";
}
- $self->add_to_syn($pombase_id, $gene_source_id, $synonym, $species_id);
+ $self->add_to_syn($pombase_id, $gene_source_id, $synonym, $species_id, $dbi);
$syn_count++;
}
}
diff --git a/misc-scripts/xref_mapping/XrefParser/RFAMParser.pm b/misc-scripts/xref_mapping/XrefParser/RFAMParser.pm
index df296026d6..868ec1ad24 100644
--- a/misc-scripts/xref_mapping/XrefParser/RFAMParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/RFAMParser.pm
@@ -33,9 +33,12 @@ sub run_script {
my ($self, $ref_arg) = @_;
my $source_id = $ref_arg->{source_id};
my $species_id = $ref_arg->{species_id};
+ my $species_name = $ref_arg->{species};
my $file = $ref_arg->{file};
my $verbose = $ref_arg->{verbose};
- my $core_db = $ref_arg->{db};
+ my $core_db = $ref_arg->{dba};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $file) ){
croak "Need to pass source_id, species_id and file as pairs";
@@ -74,8 +77,10 @@ sub run_script {
my $dba;
#get the species name
- my %id2name = $self->species_id2name;
- my $species_name = $id2name{$species_id}[0];
+ my %id2name = $self->species_id2name($dbi);
+ if (defined $species_name) { push @{$id2name{$species_id}}, $species_name; }
+ if (!defined $id2name{$species_id}) { next; }
+ $species_name = $id2name{$species_id}[0];
if ($host) {
$dba = Bio::EnsEMBL::DBSQL::DBAdaptor->new(
@@ -166,11 +171,12 @@ sub run_script {
desc => $description,
source_id => $source_id,
species_id => $species_id,
+ dbi => $dbi,
info_type => "DIRECT"} );
my @transcript_stable_ids = @{$rfam_transcript_stable_ids{$accession}};
foreach my $stable_id (@transcript_stable_ids){
- $self->add_direct_xref($xref_id, $stable_id, "Transcript", "");
+ $self->add_direct_xref($xref_id, $stable_id, "Transcript", "", $dbi);
$direct_count++;
}
$xref_count++;
diff --git a/misc-scripts/xref_mapping/XrefParser/RGDParser.pm b/misc-scripts/xref_mapping/XrefParser/RGDParser.pm
index 3a865c3e99..cfae3fea4b 100644
--- a/misc-scripts/xref_mapping/XrefParser/RGDParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/RGDParser.pm
@@ -35,6 +35,8 @@ sub run {
my $species_id = $ref_arg->{species_id};
my $files = $ref_arg->{files};
my $verbose = $ref_arg->{verbose};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $files) ){
croak "Need to pass source_id, species_id and files as pairs";
@@ -42,7 +44,7 @@ sub run {
$verbose |=0;
my $source_sql = "select source_id from source where name = 'RGD' and priority_description = 'direct_xref'";
- my $sth = $self->dbi->prepare($source_sql);
+ my $sth = $dbi->prepare($source_sql);
$sth->execute();
my ($direct_source_id);
$sth->bind_columns(\$direct_source_id);
@@ -51,9 +53,7 @@ sub run {
my $file = @{$files}[0];
- my $dbi = $self->dbi();
-
- my (%refseq) = %{$self->get_valid_codes("refseq",$species_id)};
+ my (%refseq) = %{$self->get_valid_codes("refseq",$species_id, $dbi)};
my $rgd_io = $self->get_filehandle($file);
@@ -118,6 +118,7 @@ sub run {
label => $symbol,
desc => $name,
source_id => $source_id,
+ dbi => $dbi,
species_id => $species_id} );
$count++;
my @syns = split(/\;/,$old_name);
@@ -137,9 +138,10 @@ sub run {
acc => $rgd,
label => $symbol,
desc => $name,
+ dbi => $dbi,
source_id => $direct_source_id,
species_id => $species_id} );
- my $xref_id = $self->get_xref($rgd, $direct_source_id, $species_id);
+ my $xref_id = $self->get_xref($rgd, $direct_source_id, $species_id, $dbi);
my @syns = split(/\;/, $old_name);
foreach my $syn(@syns) {
$add_syn_sth->execute($xref_id, $syn);
@@ -158,6 +160,7 @@ sub run {
desc => $name,
source_id => $source_id,
species_id => $species_id,
+ dbi => $dbi,
info_type => "MISC"} );
$mismatch++;
}
diff --git a/misc-scripts/xref_mapping/XrefParser/ReactomeParser.pm b/misc-scripts/xref_mapping/XrefParser/ReactomeParser.pm
index 522622d79a..c620073ec4 100644
--- a/misc-scripts/xref_mapping/XrefParser/ReactomeParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/ReactomeParser.pm
@@ -40,6 +40,9 @@ sub run {
my $files = $ref_arg->{files};
my $release_file = $ref_arg->{rel_file};
my $verbose = $ref_arg->{verbose};
+ my $dbi = $ref_arg->{dbi};
+ my $species_name = $ref_arg->{species};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $files) ){
croak "Needs to pass source_id, species_id and files as pairs";
@@ -64,11 +67,13 @@ sub run {
croak "Could not find release using $release_file\n";
}
- $self->set_release( $source_id, $release );
+ $self->set_release( $source_id, $release, $dbi );
}
# Create a hash of all valid names for this species
- my %species2alias = $self->species_id2name();
+ my %species2alias = $self->species_id2name($dbi);
+ if (defined $species_name) { push @{$species2alias{$species_id}}, $species_name; }
+ if (!defined $species2alias{$species_id}) { next; }
my @aliases = @{$species2alias{$species_id}};
my %alias2species_id = map {$_, 1} @aliases;
@@ -76,12 +81,11 @@ sub run {
my $err_count = 0;
my %reactome2ensembl;
- my $dbi = $self->dbi();
- my $reactome_source_id = $self->get_source_id_for_source_name("reactome", "direct");
- my $transcript_reactome_source_id = $self->get_source_id_for_source_name("reactome_transcript");
- my $gene_reactome_source_id = $self->get_source_id_for_source_name("reactome_gene");
- my $reactome_uniprot_source_id = $self->get_source_id_for_source_name("reactome", "uniprot");
+ my $reactome_source_id = $self->get_source_id_for_source_name("reactome", "direct", $dbi);
+ my $transcript_reactome_source_id = $self->get_source_id_for_source_name("reactome_transcript", undef, $dbi);
+ my $gene_reactome_source_id = $self->get_source_id_for_source_name("reactome_gene", undef, $dbi);
+ my $reactome_uniprot_source_id = $self->get_source_id_for_source_name("reactome", "uniprot", $dbi);
if($reactome_source_id < 1 || $transcript_reactome_source_id < 1 || $gene_reactome_source_id < 1){
die "Could not find source id for reactome sources???\n";
}
@@ -98,7 +102,7 @@ sub run {
print "Source_id = $reactome_uniprot_source_id\n";
}
- my (%uniprot) = %{$self->get_valid_codes("uniprot/",$species_id)};
+ my (%uniprot) = %{$self->get_valid_codes("uniprot/",$species_id, $dbi)};
my $is_uniprot = 0;
foreach my $file (@$files) {
@@ -130,6 +134,7 @@ sub run {
label => $reactome_id,
desc => $description,
source_id => $reactome_uniprot_source_id,
+ dbi => $dbi,
species_id => $species_id} );
}
$info_type = 'DEPENDENT';
@@ -158,9 +163,10 @@ sub run {
desc => $description,
info_type => $info_type,
source_id => $reactome_source_id,
+ dbi => $dbi,
species_id => $species_id} );
- $self->add_direct_xref($xref_id, $ensembl_stable_id, $type) if $type;
+ $self->add_direct_xref($xref_id, $ensembl_stable_id, $type, $dbi) if $type;
}
}
}
diff --git a/misc-scripts/xref_mapping/XrefParser/RefSeqCoordinateParser.pm b/misc-scripts/xref_mapping/XrefParser/RefSeqCoordinateParser.pm
index a49818771e..daf0b1e557 100644
--- a/misc-scripts/xref_mapping/XrefParser/RefSeqCoordinateParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/RefSeqCoordinateParser.pm
@@ -32,25 +32,28 @@ sub run_script {
my ($self, $ref_arg) = @_;
my $source_id = $ref_arg->{source_id};
my $species_id = $ref_arg->{species_id};
+ my $species_name = $ref_arg->{species};
my $file = $ref_arg->{file};
my $verbose = $ref_arg->{verbose};
my $db = $ref_arg->{dba};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $file) ){
croak "Need to pass source_id, species_id and file as pairs";
}
$verbose |=0;
- my $peptide_source_id = $self->get_source_id_for_source_name('RefSeq_peptide', 'otherfeatures');
- my $mrna_source_id = $self->get_source_id_for_source_name('RefSeq_mRNA', 'otherfeatures');
- my $ncrna_source_id = $self->get_source_id_for_source_name('RefSeq_ncRNA', 'otherfeatures');
+ my $peptide_source_id = $self->get_source_id_for_source_name('RefSeq_peptide', 'otherfeatures', $dbi);
+ my $mrna_source_id = $self->get_source_id_for_source_name('RefSeq_mRNA', 'otherfeatures', $dbi);
+ my $ncrna_source_id = $self->get_source_id_for_source_name('RefSeq_ncRNA', 'otherfeatures', $dbi);
my $pred_peptide_source_id =
- $self->get_source_id_for_source_name('RefSeq_peptide_predicted', 'otherfeatures');
+ $self->get_source_id_for_source_name('RefSeq_peptide_predicted', 'otherfeatures', $dbi);
my $pred_mrna_source_id =
- $self->get_source_id_for_source_name('RefSeq_mRNA_predicted','otherfeatures');
+ $self->get_source_id_for_source_name('RefSeq_mRNA_predicted','otherfeatures', $dbi);
my $pred_ncrna_source_id =
- $self->get_source_id_for_source_name('RefSeq_ncRNA_predicted', 'otherfeatures');
+ $self->get_source_id_for_source_name('RefSeq_ncRNA_predicted', 'otherfeatures', $dbi);
if($verbose){
print "RefSeq_peptide source ID = $peptide_source_id\n";
@@ -118,8 +121,10 @@ sub run_script {
my $registry = "Bio::EnsEMBL::Registry";
#get the species name
- my %id2name = $self->species_id2name;
- my $species_name = $id2name{$species_id}[0];
+ my %id2name = $self->species_id2name($dbi);
+ if (defined $species_name) { push @{$id2name{$species_id}}, $species_name; }
+ if (!defined $id2name{$species_id}) { next; }
+ $species_name = $id2name{$species_id}[0];
my $core_dba;
my $otherf_dba;
@@ -362,8 +367,9 @@ sub run_script {
desc => '',
source_id => $source_id,
species_id => $species_id,
+ dbi => $dbi,
info_type => 'DIRECT' });
- $self->add_direct_xref($xref_id, $best_id, "Transcript", "");
+ $self->add_direct_xref($xref_id, $best_id, "Transcript", "", $dbi);
# Also store refseq protein as direct xref for ensembl translation, if translation exists
my $ta_of = $otherf_dba->get_TranscriptAdaptor();
@@ -383,8 +389,9 @@ sub run_script {
desc => '',
source_id => $source_id,
species_id => $species_id,
+ dbi => $dbi,
info_type => 'DIRECT' });
- $self->add_direct_xref($tl_xref_id, $tl->stable_id(), "Translation", "");
+ $self->add_direct_xref($tl_xref_id, $tl->stable_id(), "Translation", "", $dbi);
}
}
}
diff --git a/misc-scripts/xref_mapping/XrefParser/RefSeqGPFFParser.pm b/misc-scripts/xref_mapping/XrefParser/RefSeqGPFFParser.pm
index c4f3834c59..71d793818a 100644
--- a/misc-scripts/xref_mapping/XrefParser/RefSeqGPFFParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/RefSeqGPFFParser.pm
@@ -42,9 +42,12 @@ sub run {
my ($self, $ref_arg) = @_;
my $source_id = $ref_arg->{source_id};
my $species_id = $ref_arg->{species_id};
+ my $species_name = $ref_arg->{species};
my $files = $ref_arg->{files};
my $release_file = $ref_arg->{rel_file};
my $verbose = $ref_arg->{verbose};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $files)){
croak "Need to pass source_id, species_id, files and rel_file as pairs";
@@ -55,21 +58,21 @@ sub run {
$peptide_source_id =
- $self->get_source_id_for_source_name('RefSeq_peptide');
+ $self->get_source_id_for_source_name('RefSeq_peptide', undef, $dbi);
$mrna_source_id =
- $self->get_source_id_for_source_name('RefSeq_mRNA','refseq');
+ $self->get_source_id_for_source_name('RefSeq_mRNA','refseq', $dbi);
$ncrna_source_id =
- $self->get_source_id_for_source_name('RefSeq_ncRNA');
+ $self->get_source_id_for_source_name('RefSeq_ncRNA', undef, $dbi);
$pred_peptide_source_id =
- $self->get_source_id_for_source_name('RefSeq_peptide_predicted');
+ $self->get_source_id_for_source_name('RefSeq_peptide_predicted', undef, $dbi);
$pred_mrna_source_id =
- $self->get_source_id_for_source_name('RefSeq_mRNA_predicted','refseq');
+ $self->get_source_id_for_source_name('RefSeq_mRNA_predicted','refseq', $dbi);
$pred_ncrna_source_id =
- $self->get_source_id_for_source_name('RefSeq_ncRNA_predicted');
+ $self->get_source_id_for_source_name('RefSeq_ncRNA_predicted', undef, $dbi);
- $entrez_source_id = $self->get_source_id_for_source_name('EntrezGene');
- $wiki_source_id = $self->get_source_id_for_source_name('WikiGene');
+ $entrez_source_id = $self->get_source_id_for_source_name('EntrezGene', undef, $dbi);
+ $wiki_source_id = $self->get_source_id_for_source_name('WikiGene', undef, $dbi);
if($verbose){
print "RefSeq_peptide source ID = $peptide_source_id\n";
@@ -80,17 +83,17 @@ sub run {
print "RefSeq_ncRNA_predicted source ID = $pred_ncrna_source_id\n" ;
}
- (%entrez) = %{$self->get_acc_to_label("EntrezGene",$species_id)};
+ (%entrez) = %{$self->get_acc_to_label("EntrezGene",$species_id, undef, $dbi)};
my @xrefs;
foreach my $file (@files) {
my $xrefs =
- $self->create_xrefs( $file, $species_id, $verbose );
+ $self->create_xrefs( $file, $species_id, $verbose, $dbi, $species_name );
if ( !defined( $xrefs ) ) {
return 1; #error
}
- $self->upload_xref_object_graphs( $xrefs )
+ $self->upload_xref_object_graphs( $xrefs, $dbi )
}
if ( defined $release_file ) {
@@ -107,13 +110,13 @@ sub run {
print "RefSeq release: '$release'\n" if($verbose);
- $self->set_release( $source_id, $release );
- $self->set_release( $peptide_source_id, $release );
- $self->set_release( $mrna_source_id, $release );
- $self->set_release( $ncrna_source_id, $release );
- $self->set_release( $pred_mrna_source_id, $release );
- $self->set_release( $pred_ncrna_source_id, $release );
- $self->set_release( $pred_peptide_source_id, $release );
+ $self->set_release( $source_id, $release, $dbi );
+ $self->set_release( $peptide_source_id, $release, $dbi );
+ $self->set_release( $mrna_source_id, $release, $dbi );
+ $self->set_release( $ncrna_source_id, $release, $dbi );
+ $self->set_release( $pred_mrna_source_id, $release, $dbi );
+ $self->set_release( $pred_ncrna_source_id, $release, $dbi );
+ $self->set_release( $pred_peptide_source_id, $release, $dbi );
}
return 0; # successful
@@ -127,26 +130,28 @@ sub run {
# Slightly different formats
sub create_xrefs {
- my ($self, $file,$species_id, $verbose ) = @_;
+ my ($self, $file,$species_id, $verbose, $dbi, $species_name ) = @_;
# Create a hash of all valid names and taxon_ids for this species
- my %species2name = $self->species_id2name();
- my %species2tax = $self->species_id2taxonomy();
+ my %species2name = $self->species_id2name($dbi);
+ if (defined $species_name) { push @{$species2name{$species_id}}, $species_name; }
+ if (!defined $species2name{$species_id}) { next; }
+ my %species2tax = $self->species_id2taxonomy($dbi);
+ push @{$species2tax{$species_id}}, $species_id;
my @names = @{$species2name{$species_id}};
my @tax_ids = @{$species2tax{$species_id}};
my %name2species_id = map{ $_=>$species_id } @names;
my %taxonomy2species_id = map{ $_=>$species_id } @tax_ids;
# Retrieve existing RefSeq mRNA
- my (%refseq_ids) = (%{ $self->get_valid_codes("RefSeq_mRNA", $species_id) }, %{ $self->get_valid_codes("RefSeq_mRNA_predicted", $species_id) });
- my (%entrez_ids) = %{ $self->get_valid_codes("EntrezGene", $species_id) };
- my (%wiki_ids) = %{ $self->get_valid_codes("WikiGene", $species_id) };
+ my (%refseq_ids) = (%{ $self->get_valid_codes("RefSeq_mRNA", $species_id, $dbi) }, %{ $self->get_valid_codes("RefSeq_mRNA_predicted", $species_id, $dbi) });
+ my (%entrez_ids) = %{ $self->get_valid_codes("EntrezGene", $species_id, $dbi) };
+ my (%wiki_ids) = %{ $self->get_valid_codes("WikiGene", $species_id, $dbi) };
- my %dependent_sources = $self->get_xref_sources();
+ my %dependent_sources = $self->get_xref_sources($dbi);
- my $dbi = $self->dbi();
- my $add_dependent_xref_sth = $self->dbi->prepare("INSERT INTO dependent_xref (master_xref_id,dependent_xref_id, linkage_source_id) VALUES (?,?, $entrez_source_id)");
+ my $add_dependent_xref_sth = $dbi->prepare("INSERT INTO dependent_xref (master_xref_id,dependent_xref_id, linkage_source_id) VALUES (?,?, $entrez_source_id)");
my $refseq_io = $self->get_filehandle($file);
diff --git a/misc-scripts/xref_mapping/XrefParser/RefSeqParser.pm b/misc-scripts/xref_mapping/XrefParser/RefSeqParser.pm
index d3a3582fb3..896005cfc7 100644
--- a/misc-scripts/xref_mapping/XrefParser/RefSeqParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/RefSeqParser.pm
@@ -37,6 +37,8 @@ sub run {
my $files = $ref_arg->{files};
my $release_file = $ref_arg->{rel_file};
my $verbose = $ref_arg->{verbose};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $files) ){
croak "Need to pass source_id, species_id and files as pairs";
@@ -47,18 +49,18 @@ sub run {
my $peptide_source_id =
- $self->get_source_id_for_source_name('RefSeq_peptide');
+ $self->get_source_id_for_source_name('RefSeq_peptide', undef, $dbi);
my $mrna_source_id =
- $self->get_source_id_for_source_name('RefSeq_mRNA','refseq');
+ $self->get_source_id_for_source_name('RefSeq_mRNA', undef, $dbi);
my $ncrna_source_id =
- $self->get_source_id_for_source_name('RefSeq_ncRNA');
+ $self->get_source_id_for_source_name('RefSeq_ncRNA', undef, $dbi);
my $pred_peptide_source_id =
- $self->get_source_id_for_source_name('RefSeq_peptide_predicted');
+ $self->get_source_id_for_source_name('RefSeq_peptide_predicted', undef, $dbi);
my $pred_mrna_source_id =
- $self->get_source_id_for_source_name('RefSeq_mRNA_predicted','refseq');
+ $self->get_source_id_for_source_name('RefSeq_mRNA_predicted','refseq', $dbi);
my $pred_ncrna_source_id =
- $self->get_source_id_for_source_name('RefSeq_ncRNA_predicted');
+ $self->get_source_id_for_source_name('RefSeq_ncRNA_predicted', undef, $dbi);
if($verbose){
print "RefSeq_peptide source ID = $peptide_source_id\n";
@@ -78,7 +80,7 @@ sub run {
$mrna_source_id, $ncrna_source_id,
$pred_mrna_source_id, $pred_ncrna_source_id,
$file,
- $species_id );
+ $species_id, $dbi );
if ( !defined($xrefs) ) {
return 1; #error
@@ -88,7 +90,7 @@ sub run {
push @xrefs, @{$xrefs};
}
- if ( !defined( $self->upload_xref_object_graphs( \@xrefs ) ) ) {
+ if ( !defined( $self->upload_xref_object_graphs( \@xrefs, $dbi ) ) ) {
return 1; # error
}
@@ -107,13 +109,13 @@ sub run {
print "RefSeq release: '$release'\n";
- $self->set_release( $source_id, $release );
- $self->set_release( $peptide_source_id, $release );
- $self->set_release( $mrna_source_id, $release );
- $self->set_release( $ncrna_source_id, $release );
- $self->set_release( $pred_peptide_source_id, $release );
- $self->set_release( $pred_mrna_source_id, $release );
- $self->set_release( $pred_ncrna_source_id, $release );
+ $self->set_release( $source_id, $release, $dbi );
+ $self->set_release( $peptide_source_id, $release, $dbi );
+ $self->set_release( $mrna_source_id, $release, $dbi );
+ $self->set_release( $ncrna_source_id, $release, $dbi );
+ $self->set_release( $pred_peptide_source_id, $release, $dbi );
+ $self->set_release( $pred_mrna_source_id, $release, $dbi );
+ $self->set_release( $pred_ncrna_source_id, $release, $dbi );
}
return 0; # successfull
@@ -130,10 +132,10 @@ sub run {
sub create_xrefs {
my ($self, $peptide_source_id, $pred_peptide_source_id,
$mrna_source_id, $ncrna_source_id,
- $pred_mrna_source_id, $pred_ncrna_source_id, $file, $species_id ) = @_;
+ $pred_mrna_source_id, $pred_ncrna_source_id, $file, $species_id, $dbi ) = @_;
# Create a hash of all valid names for this species
- my %species2name = $self->species_id2name();
+ my %species2name = $self->species_id2name($dbi);
my @names = @{$species2name{$species_id}};
my %name2species_id = map{ $_=>$species_id } @names;
# my %name2species_id = $self->name2species_id();
diff --git a/misc-scripts/xref_mapping/XrefParser/RefSeq_CCDSParser.pm b/misc-scripts/xref_mapping/XrefParser/RefSeq_CCDSParser.pm
index 1662887bf7..bbca7c6f9c 100644
--- a/misc-scripts/xref_mapping/XrefParser/RefSeq_CCDSParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/RefSeq_CCDSParser.pm
@@ -35,6 +35,8 @@ sub run_script {
my $species_id = $ref_arg->{species_id};
my $file = $ref_arg->{file};
my $verbose = $ref_arg->{verbose};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $file) ){
croak "Need to pass source_id, species_id and file as pairs";
@@ -61,9 +63,9 @@ sub run_script {
}
my $mrna_source_id =
- $self->get_source_id_for_source_name('RefSeq_mRNA','ccds');
+ $self->get_source_id_for_source_name('RefSeq_mRNA','ccds', $dbi);
my $pred_mrna_source_id =
- $self->get_source_id_for_source_name('RefSeq_mRNA_predicted','ccds');
+ $self->get_source_id_for_source_name('RefSeq_mRNA_predicted','ccds', $dbi);
if($verbose){
print "RefSeq_mRNA source ID = $mrna_source_id\n";
@@ -77,8 +79,6 @@ sub run_script {
my %version;
my %description;
- my $dbi = $self->dbi();
-
my $sql =(<<'RSS');
SELECT xref.accession, xref.label, xref.version, xref.description
FROM xref, source
@@ -221,11 +221,12 @@ CCDS
desc => $description{$refseq},
source_id => $new_source_id,
species_id => $species_id,
+ dbi => $dbi,
info_type => "DIRECT"} );
foreach my $stable_id (@{$internal_to_stable_id{$internal_id}}){
- $self->add_direct_xref($xref_id, $stable_id, "Transcript", "");
+ $self->add_direct_xref($xref_id, $stable_id, "Transcript", "", $dbi);
$direct_count++;
}
diff --git a/misc-scripts/xref_mapping/XrefParser/SGDParser.pm b/misc-scripts/xref_mapping/XrefParser/SGDParser.pm
index 12d2d339a0..a68964b359 100644
--- a/misc-scripts/xref_mapping/XrefParser/SGDParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/SGDParser.pm
@@ -37,6 +37,8 @@ sub run {
my $species_id = $ref_arg->{species_id};
my $files = $ref_arg->{files};
my $verbose = $ref_arg->{verbose};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $files) ){
croak "Need to pass source_id, species_id and files as pairs";
@@ -45,9 +47,9 @@ sub run {
my $file = @{$files}[0];
- my $gene_source_id = $self->get_source_id_for_source_name("SGD_GENE");
+ my $gene_source_id = $self->get_source_id_for_source_name("SGD_GENE", undef, $dbi);
#my $transcript_source_id = $self->get_source_id_for_source_name("SGD_TRANSCRIPT");
- my $translation_source_id = $self->get_source_id_for_source_name("SGD_TRANSLATION");
+ my $translation_source_id = $self->get_source_id_for_source_name("SGD_TRANSLATION", undef, $dbi);
my $sgd_io = $self->get_filehandle($file);
@@ -97,16 +99,18 @@ sub run {
desc => $desc,
source_id => $gene_source_id,
species_id => $species_id,
+ dbi => $dbi,
info_type => "DIRECT"} );
- $self->add_direct_xref($gene_xref_id, $orf_name, "Gene", "DIRECT");
+ $self->add_direct_xref($gene_xref_id, $orf_name, "Gene", "DIRECT", $dbi);
my $translation_xref_id = $self->add_xref({ acc => $sgd_id,
label => $locus_name,
desc => $desc,
source_id => $translation_source_id,
species_id => $species_id,
+ dbi => $dbi,
info_type => "DIRECT"} );
- $self->add_direct_xref($translation_xref_id, $orf_name, "Translation", "DIRECT");
+ $self->add_direct_xref($translation_xref_id, $orf_name, "Translation", "DIRECT", $dbi);
$xref_count++;
@@ -114,7 +118,7 @@ sub run {
if ($verbose) {
# print STDERR "adding synonym, $synonym\n";
}
- $self->add_to_syn($sgd_id, $gene_source_id, $synonym, $species_id);
+ $self->add_to_syn($sgd_id, $gene_source_id, $synonym, $species_id, $dbi);
$syn_count++;
}
diff --git a/misc-scripts/xref_mapping/XrefParser/UCSCParser.pm b/misc-scripts/xref_mapping/XrefParser/UCSCParser.pm
index df051a8623..e8f1af9a01 100644
--- a/misc-scripts/xref_mapping/XrefParser/UCSCParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/UCSCParser.pm
@@ -33,6 +33,8 @@ sub run {
my $species_id = $ref_arg->{species_id};
my $files = $ref_arg->{files};
my $verbose = $ref_arg->{verbose};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $files) ){
croak "Need to pass source_id, species_id and files as pairs";
@@ -42,7 +44,7 @@ sub run {
my $data_file = @{$files}[0];
# Get the $source_id for the "UCSC" source.
- $source_id = $self->get_source_id_for_source_name('UCSC');
+ $source_id = $self->get_source_id_for_source_name('UCSC', undef, $dbi);
my $data_io = $self->get_filehandle($data_file);
@@ -104,6 +106,7 @@ sub run {
'cdsStart' => $cdsStart,
'cdsEnd' => $cdsEnd,
'exonStarts' => $exonStarts,
+ 'dbi' => $dbi,
'exonEnds' => $exonEnds );
$self->add_xref( $source_id, $species_id, \%xref );
diff --git a/misc-scripts/xref_mapping/XrefParser/UniProtDirectParser.pm b/misc-scripts/xref_mapping/XrefParser/UniProtDirectParser.pm
index 0e042d70e2..916a887671 100644
--- a/misc-scripts/xref_mapping/XrefParser/UniProtDirectParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/UniProtDirectParser.pm
@@ -40,6 +40,8 @@ sub run_script {
my $species_id = $ref_arg->{species_id};
my $file = $ref_arg->{file};
my $verbose = $ref_arg->{verbose};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $file) ){
croak "Need to pass source_id, species_id and file as pairs";
@@ -111,9 +113,8 @@ sub run_script {
push @{$prot2ensembl{$prot}}, $ens;
}
}
- my $dbi = $self->dbi();
- my $sw_source_id = $self->get_source_id_for_source_name("uniprot/swissprot","sequence_mapped");
+ my $sw_source_id = $self->get_source_id_for_source_name("uniprot/swissprot","sequence_mapped", $dbi);
if($sw_source_id < 1){
die "Could not find source id for uniprot/swissprot ???\n";
}
@@ -174,6 +175,7 @@ sub run_script {
desc => $description,
source_id => $source_id,
species_id => $species_id,
+ dbi => $dbi,
info_type => "DIRECT"} );
@@ -193,7 +195,7 @@ sub run_script {
#add the direct xref entry
#
- $self->add_direct_xref( $xref_id, $trans, "Translation", '');
+ $self->add_direct_xref( $xref_id, $trans, "Translation", '', $dbi);
# print ":".$trans;
#
diff --git a/misc-scripts/xref_mapping/XrefParser/UniProtParser.pm b/misc-scripts/xref_mapping/XrefParser/UniProtParser.pm
index 0086f30049..336a1cb583 100644
--- a/misc-scripts/xref_mapping/XrefParser/UniProtParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/UniProtParser.pm
@@ -43,9 +43,12 @@ sub run {
my ($self, $ref_arg) = @_;
my $source_id = $ref_arg->{source_id};
my $species_id = $ref_arg->{species_id};
+ my $species_name = $ref_arg->{species};
my $files = $ref_arg->{files};
my $release_file = $ref_arg->{rel_file};
my $verbose = $ref_arg->{verbose};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $files)){
croak "Need to pass source_id, species_id, files and rel_file as pairs";
@@ -57,15 +60,15 @@ sub run {
my ( $sp_source_id, $sptr_source_id, $sp_release, $sptr_release, $sptr_non_display_source_id, $sp_direct_source_id, $sptr_direct_source_id );
$sp_source_id =
- $self->get_source_id_for_source_name('Uniprot/SWISSPROT','sequence_mapped');
+ $self->get_source_id_for_source_name('Uniprot/SWISSPROT','sequence_mapped', $dbi);
$sptr_source_id =
- $self->get_source_id_for_source_name('Uniprot/SPTREMBL', 'sequence_mapped');
+ $self->get_source_id_for_source_name('Uniprot/SPTREMBL', 'sequence_mapped', $dbi);
$sptr_non_display_source_id =
- $self->get_source_id_for_source_name('Uniprot/SPTREMBL', 'protein_evidence_gt_2');
+ $self->get_source_id_for_source_name('Uniprot/SPTREMBL', 'protein_evidence_gt_2', $dbi);
- $sp_direct_source_id = $self->get_source_id_for_source_name('Uniprot/SWISSPROT', 'direct');
- $sptr_direct_source_id = $self->get_source_id_for_source_name('Uniprot/SPTREMBL', 'direct');
+ $sp_direct_source_id = $self->get_source_id_for_source_name('Uniprot/SWISSPROT', 'direct', $dbi);
+ $sptr_direct_source_id = $self->get_source_id_for_source_name('Uniprot/SPTREMBL', 'direct', $dbi);
print "SwissProt source id for $file: $sp_source_id\n" if ($verbose);
print "SpTREMBL source id for $file: $sptr_source_id\n" if ($verbose);
@@ -74,7 +77,7 @@ sub run {
print "SpTREMBL direct source id for $file: $sptr_direct_source_id\n" if ($verbose);
$self->create_xrefs( $sp_source_id, $sptr_source_id, $sptr_non_display_source_id, $species_id,
- $file, $verbose, $sp_direct_source_id, $sptr_direct_source_id );
+ $file, $verbose, $sp_direct_source_id, $sptr_direct_source_id, $dbi );
if ( defined $release_file ) {
# Parse Swiss-Prot and SpTrEMBL release info from
@@ -92,11 +95,11 @@ sub run {
$release_io->close();
# Set releases
- $self->set_release( $sp_source_id, $sp_release );
- $self->set_release( $sptr_source_id, $sptr_release );
- $self->set_release( $sptr_non_display_source_id, $sptr_release );
- $self->set_release( $sp_direct_source_id, $sp_release );
- $self->set_release( $sptr_direct_source_id,$sptr_release );
+ $self->set_release( $sp_source_id, $sp_release, $dbi );
+ $self->set_release( $sptr_source_id, $sptr_release, $dbi );
+ $self->set_release( $sptr_non_display_source_id, $sptr_release, $dbi );
+ $self->set_release( $sp_direct_source_id, $sp_release, $dbi );
+ $self->set_release( $sptr_direct_source_id,$sptr_release, $dbi );
}
@@ -108,7 +111,7 @@ sub run {
# Parse file into array of xref objects
sub create_xrefs {
- my ($self, $sp_source_id, $sptr_source_id, $sptr_non_display_source_id, $species_id, $file, $verbose, $sp_direct_source_id, $sptr_direct_source_id ) = @_;
+ my ($self, $sp_source_id, $sptr_source_id, $sptr_non_display_source_id, $species_id, $file, $verbose, $sp_direct_source_id, $sptr_direct_source_id, $dbi ) = @_;
my $num_sp = 0;
my $num_sptr = 0;
@@ -118,16 +121,16 @@ sub create_xrefs {
my $num_direct_sp = 0;
my $num_direct_sptr = 0;
- my %dependent_sources = $self->get_xref_sources();
+ my %dependent_sources = $self->get_xref_sources($dbi);
if(defined($dependent_sources{'MGI'})){
- $dependent_sources{'MGI'} = $self->get_source_id_for_source_name("MGI","uniprot");
+ $dependent_sources{'MGI'} = $self->get_source_id_for_source_name("MGI","uniprot", $dbi);
}
my (%genemap) =
- %{ $self->get_valid_codes( "mim_gene", $species_id ) };
+ %{ $self->get_valid_codes( "mim_gene", $species_id, $dbi ) };
my (%morbidmap) =
- %{ $self->get_valid_codes( "mim_morbid", $species_id ) };
+ %{ $self->get_valid_codes( "mim_morbid", $species_id, $dbi ) };
my $uniprot_io = $self->get_filehandle($file);
if ( !defined $uniprot_io ) { return }
@@ -137,7 +140,8 @@ sub create_xrefs {
local $/ = "//\n";
# Create a hash of all valid taxon_ids for this species
- my %species2tax = $self->species_id2taxonomy();
+ my %species2tax = $self->species_id2taxonomy($dbi);
+ push @{$species2tax{$species_id}}, $species_id;
my @tax_ids = @{$species2tax{$species_id}};
my %taxonomy2species_id = map{ $_=>$species_id } @tax_ids;
@@ -150,7 +154,6 @@ sub create_xrefs {
my %mgi_label_to_desc;
my %mgi_label_to_acc;
- my $dbi = $self->dbi();
my $sth = $dbi->prepare("SELECT x.accession, x.label, x.description from xref x, source s where x.source_id = s.source_id and s.name like 'MGI' and s.priority_description like 'descriptions'");
$sth->execute() or croak( $dbi->errstr() );
@@ -552,14 +555,14 @@ sub create_xrefs {
push @xrefs, $xref;
if ($count > 1000) {
- $self->upload_xref_object_graphs(\@xrefs);
+ $self->upload_xref_object_graphs(\@xrefs, $dbi);
$count = 0;
undef @xrefs;
}
}
- $self->upload_xref_object_graphs(\@xrefs) if scalar(@xrefs) > 0;
+ $self->upload_xref_object_graphs(\@xrefs, $dbi) if scalar(@xrefs) > 0;
$uniprot_io->close();
diff --git a/misc-scripts/xref_mapping/XrefParser/UniProtParser_descriptions_only.pm b/misc-scripts/xref_mapping/XrefParser/UniProtParser_descriptions_only.pm
index 850d2f580b..1d3db073a7 100644
--- a/misc-scripts/xref_mapping/XrefParser/UniProtParser_descriptions_only.pm
+++ b/misc-scripts/xref_mapping/XrefParser/UniProtParser_descriptions_only.pm
@@ -48,6 +48,8 @@ sub run {
my $files = $ref_arg->{files};
my $release_file = $ref_arg->{rel_file};
my $verbose = $ref_arg->{verbose};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $files) or (!defined $release_file)){
croak "Need to pass source_id, species_id, files and rel_file as pairs";
@@ -59,16 +61,16 @@ sub run {
my ( $sp_source_id, $sptr_source_id, $sp_release, $sptr_release );
$sp_source_id =
- $self->get_source_id_for_source_name('Uniprot/SWISSPROT',"sequence_mapped");
+ $self->get_source_id_for_source_name('Uniprot/SWISSPROT',"sequence_mapped", $dbi);
$sptr_source_id =
- $self->get_source_id_for_source_name('Uniprot/SPTREMBL');
+ $self->get_source_id_for_source_name('Uniprot/SPTREMBL', undef, $dbi);
print "SwissProt source id for $file: $sp_source_id\n" if($verbose);
print "SpTREMBL source id for $file: $sptr_source_id\n" if($verbose);
my @xrefs =
- $self->create_xrefs( $sp_source_id, $sptr_source_id, $species_id, $file, $verbose );
+ $self->create_xrefs( $sp_source_id, $sptr_source_id, $species_id, $file, $verbose, $dbi );
if ( !@xrefs ) {
return 1; # 1 error
@@ -81,7 +83,7 @@ sub run {
# }
# upload
- if(!defined($self->upload_xref_object_graphs(@xrefs))){
+ if(!defined($self->upload_xref_object_graphs(@xrefs, $dbi))){
return 1;
}
@@ -90,10 +92,10 @@ sub run {
# below...
my $sp_pred_source_id =
$self->get_source_id_for_source_name(
- 'Uniprot/SWISSPROT_predicted');
+ 'Uniprot/SWISSPROT_predicted', undef, $dbi);
my $sptr_pred_source_id =
$self->get_source_id_for_source_name(
- 'Uniprot/SPTREMBL_predicted');
+ 'Uniprot/SPTREMBL_predicted', undef, $dbi);
# Parse Swiss-Prot and SpTrEMBL release info from
# $release_file.
@@ -110,10 +112,10 @@ sub run {
$release_io->close();
# Set releases
- $self->set_release( $sp_source_id, $sp_release );
- $self->set_release( $sptr_source_id, $sptr_release );
- $self->set_release( $sp_pred_source_id, $sp_release );
- $self->set_release( $sptr_pred_source_id, $sptr_release );
+ $self->set_release( $sp_source_id, $sp_release, $dbi );
+ $self->set_release( $sptr_source_id, $sptr_release, $dbi );
+ $self->set_release( $sp_pred_source_id, $sp_release, $dbi );
+ $self->set_release( $sptr_pred_source_id, $sptr_release, $dbi );
}
@@ -125,7 +127,7 @@ sub run {
# Parse file into array of xref objects
sub create_xrefs {
- my ($self, $sp_source_id, $sptr_source_id, $species_id, $file, $verbose ) = @_;
+ my ($self, $sp_source_id, $sptr_source_id, $species_id, $file, $verbose, $dbi ) = @_;
my $num_sp = 0;
my $num_sptr = 0;
@@ -135,9 +137,9 @@ sub create_xrefs {
# Get predicted equivalents of various sources used here
my $sp_pred_source_id =
- $self->get_source_id_for_source_name('Uniprot/SWISSPROT_predicted');
+ $self->get_source_id_for_source_name('Uniprot/SWISSPROT_predicted', undef, $dbi);
my $sptr_pred_source_id =
- $self->get_source_id_for_source_name('Uniprot/SPTREMBL_predicted');
+ $self->get_source_id_for_source_name('Uniprot/SPTREMBL_predicted', undef, $dbi);
print "Predicted SwissProt source id for $file: $sp_pred_source_id\n" if($verbose);
print "Predicted SpTREMBL source id for $file: $sptr_pred_source_id\n" if($verbose);
@@ -150,7 +152,7 @@ sub create_xrefs {
local $/ = "//\n";
# Create a hash of all valid taxon_ids for this species
- my %species2tax = $self->species_id2taxonomy();
+ my %species2tax = $self->species_id2taxonomy($dbi);
my @tax_ids = @{$species2tax{$species_id}};
my %taxonomy2species_id = map{ $_=>$species_id } @tax_ids;
diff --git a/misc-scripts/xref_mapping/XrefParser/VGNCParser.pm b/misc-scripts/xref_mapping/XrefParser/VGNCParser.pm
index cd3f395622..346ad87a9b 100644
--- a/misc-scripts/xref_mapping/XrefParser/VGNCParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/VGNCParser.pm
@@ -32,6 +32,8 @@ sub run {
my $species_id = $ref_arg->{species_id};
my $files = $ref_arg->{files};
my $verbose = $ref_arg->{verbose};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $files) ){
croak "Need to pass source_id, species_id, files and rel_file as pairs";
@@ -49,9 +51,9 @@ sub run {
return 1;
}
- my $source_name = $self->get_source_name_for_source_id($source_id);
+ my $source_name = $self->get_source_name_for_source_id($source_id, $dbi);
# Create a hash of all valid taxon_ids for this species
- my %species2tax = $self->species_id2taxonomy();
+ my %species2tax = $self->species_id2taxonomy($dbi);
my @tax_ids = @{$species2tax{$species_id}};
my %taxonomy2species_id = map{ $_=>$species_id } @tax_ids;
@@ -80,13 +82,15 @@ sub run {
type => 'gene',
acc => $acc,
label => $symbol,
- desc => $name,,
+ desc => $name,
+ dbi => $dbi,
source_id => $source_id,
species_id => $species_id} );
$self->add_synonyms_for_hgnc( {source_id => $source_id,
name => $acc,
species_id => $species_id,
+ dbi => $dbi,
dead => $previous_symbols,
alias => $synonyms});
diff --git a/misc-scripts/xref_mapping/XrefParser/XenopusJamboreeParser.pm b/misc-scripts/xref_mapping/XrefParser/XenopusJamboreeParser.pm
index bcdd99873a..647edb3c43 100644
--- a/misc-scripts/xref_mapping/XrefParser/XenopusJamboreeParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/XenopusJamboreeParser.pm
@@ -36,6 +36,8 @@ sub run {
my $species_id = $ref_arg->{species_id};
my $files = $ref_arg->{files};
my $verbose = $ref_arg->{verbose};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $files) ){
croak "Need to pass source_id, species_id and files as pairs";
@@ -64,6 +66,7 @@ sub run {
acc => $acc,
label => $label,
desc => $desc,
+ dbi => $dbi,
source_id => $source_id,
species_id => $species_id });
$count++;
diff --git a/misc-scripts/xref_mapping/XrefParser/ZFINParser.pm b/misc-scripts/xref_mapping/XrefParser/ZFINParser.pm
index 7fb0d7da46..465716463a 100644
--- a/misc-scripts/xref_mapping/XrefParser/ZFINParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/ZFINParser.pm
@@ -34,6 +34,8 @@ sub run {
my $species_id = $ref_arg->{species_id};
my $files = $ref_arg->{files};
my $verbose = $ref_arg->{verbose};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $files) ){
croak "Need to pass source_id, species_id and files as pairs";
@@ -43,8 +45,8 @@ sub run {
my $file = @{$files}[0];
my $dir = dirname($file);
- my (%swiss) = %{$self->get_valid_codes("uniprot/",$species_id)};
- my (%refseq) = %{$self->get_valid_codes("refseq",$species_id)};
+ my (%swiss) = %{$self->get_valid_codes("uniprot/",$species_id, $dbi)};
+ my (%refseq) = %{$self->get_valid_codes("refseq",$species_id, $dbi)};
my $swissprot_io =
$self->get_filehandle( catfile( $dir, 'uniprot.txt' ) );
@@ -62,9 +64,6 @@ sub run {
my %description;
- my $dbi = $self->dbi();
-
-
my $sql = "insert ignore into synonym (xref_id, synonym) values (?, ?)";
my $add_syn_sth = $dbi->prepare($sql);
@@ -109,6 +108,7 @@ sub run {
label => $label,
desc => $description{$zfin},
source_id => $source_id,
+ dbi => $dbi,
species_id => $species_id} );
$spcount++;
}
@@ -143,6 +143,7 @@ sub run {
label => $label,
desc => $description{$zfin},
source_id => $source_id,
+ dbi => $dbi,
species_id => $species_id} );
$rscount++;
}
@@ -154,7 +155,7 @@ sub run {
$refseq_io->close();
- my (%zfin) = %{$self->get_valid_codes("zfin",$species_id)};
+ my (%zfin) = %{$self->get_valid_codes("zfin",$species_id, $dbi)};
my $zfin_io = $self->get_filehandle( catfile( $dir, 'aliases.txt' ) );
@@ -183,7 +184,7 @@ sub run {
chomp;
my ($acc, undef, undef, $syn) = split (/\t/,$_);
if(defined($zfin{$acc})){
- $self->add_to_syn_for_mult_sources($acc, $sources, $syn, $species_id);
+ $self->add_to_syn_for_mult_sources($acc, $sources, $syn, $species_id, $dbi);
$syncount++;
}
}
diff --git a/misc-scripts/xref_mapping/XrefParser/miRBaseParser.pm b/misc-scripts/xref_mapping/XrefParser/miRBaseParser.pm
index 8f8149ea54..28b98b5344 100644
--- a/misc-scripts/xref_mapping/XrefParser/miRBaseParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/miRBaseParser.pm
@@ -32,8 +32,11 @@ sub run {
my ($self, $ref_arg) = @_;
my $source_id = $ref_arg->{source_id};
my $species_id = $ref_arg->{species_id};
+ my $species_name = $ref_arg->{species};
my $files = $ref_arg->{files};
my $verbose = $ref_arg->{verbose};
+ my $dbi = $ref_arg->{dbi};
+ $dbi = $self->dbi unless defined $dbi;
if((!defined $source_id) or (!defined $species_id) or (!defined $files) ){
croak "Need to pass source_id, species_id and files as pairs";
@@ -46,12 +49,12 @@ sub run {
$species_id = $self->get_species_id_for_filename($file);
}
- my $xrefs = $self->create_xrefs($source_id, $file, $species_id);
+ my $xrefs = $self->create_xrefs($source_id, $file, $species_id, $dbi, $species_name);
if(!defined($xrefs)){
return 1; #error
}
# upload
- if(!defined($self->upload_xref_object_graphs($xrefs))){
+ if(!defined($self->upload_xref_object_graphs($xrefs, $dbi))){
return 1;
}
return 0; # successfull
@@ -63,9 +66,11 @@ sub run {
sub create_xrefs {
- my ($self, $source_id, $file, $species_id) = @_;
+ my ($self, $source_id, $file, $species_id, $dbi, $species_name) = @_;
- my %species2name = $self->species_id2name();
+ my %species2name = $self->species_id2name($dbi);
+ if (defined $species_name) { push @{$species2name{$species_id}}, $species_name; }
+ if (!defined $species2name{$species_id}) { next; }
my @names = @{$species2name{$species_id}};
my %name2species_id = map{ $_=>$species_id } @names;
--
GitLab