Skip to content
Snippets Groups Projects
Commit b0b66b4d authored by Ian Longden's avatar Ian Longden
Browse files

source_id removed from synonym,dependent_xref and primary_xref this... 

source_id removed from synonym,dependent_xref and primary_xref this information is available via the xref. linkage_source_id added to dependent xref. create_xrefs now also returns a hashref
parent 620200c3
No related branches found
No related tags found
No related merge requests found
Hide whitespace changes
Inline Side-by-side
misc-scripts/xref_mapping/XrefParser/BaseParser.pm
+ 26
− 20
View file @ b0b66b4d
......@@ -278,21 +278,24 @@ sub get_valid_codes{
sub upload_xref_object_graphs {
my ($self, @xrefs) = @_;
my ($self, $rxrefs) = @_;
my $dbi = dbi();
if ($#xrefs > -1) {
# print "count = ".$#$rxrefs."\n";
if ($#$rxrefs > -1) {
# remove all existing xrefs with same source ID(s)
delete_by_source(\@xrefs);
delete_by_source($rxrefs);
# upload new ones
print "Uploading xrefs\n";
my $xref_sth = $dbi->prepare("INSERT INTO xref (accession,version,label,description,source_id,species_id) VALUES(?,?,?,?,?,?)");
my $pri_insert_sth = $dbi->prepare("INSERT INTO primary_xref 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=?");
my $syn_sth = $dbi->prepare("INSERT INTO synonym VALUES(?,?,?)");
my $syn_sth = $dbi->prepare("INSERT INTO synonym VALUES(?,?)");
my $dep_sth = $dbi->prepare("INSERT INTO dependent_xref VALUES(?,?,?,?)");
my $xref_update_label_sth = $dbi->prepare("UPDATE xref SET label=? WHERE xref_id=?");
my $xref_update_descr_sth = $dbi->prepare("UPDATE xref SET description=? WHERE xref_id=?");
......@@ -300,7 +303,7 @@ sub upload_xref_object_graphs {
local $xref_sth->{RaiseError}; # disable error handling here as we'll do it ourselves
local $xref_sth->{PrintError};
foreach my $xref (@xrefs) {
foreach my $xref (@{$rxrefs}) {
my $xref_id;
# Create entry in xref table and note ID
if(! $xref_sth->execute($xref->{ACCESSION},
......@@ -332,23 +335,26 @@ sub upload_xref_object_graphs {
$pri_insert_sth->execute($xref_id,
$xref->{SEQUENCE},
$xref->{SEQUENCE_TYPE},
$xref->{STATUS},
$xref->{SOURCE_ID}) || die $dbi->errstr;
$xref->{STATUS}) || die $dbi->errstr;
}
# if there are synonyms, create xrefs for them and entries in the synonym table
foreach my $syn (@{$xref->{SYNONYMS}}) {
$xref_sth->execute($syn,
"",
"",
"",
$xref->{SOURCE_ID},
$xref->{SPECIES_ID});
my $syn_xref_id = insert_or_select($xref_sth, $dbi->err, $syn, $xref->{SOURCE_ID});
$syn_sth->execute($xref_id, $syn_xref_id, $xref->{SOURCE_ID} ) || die $dbi->errstr;
my $syn_xref_id = get_xref($syn,$xref->{SOURCE_ID});
if(!defined($syn_xref_id)){
$xref_sth->execute($syn,
"",
"",
"",
$xref->{SOURCE_ID},
$xref->{SPECIES_ID});
$syn_xref_id = get_xref($syn,$xref->{SOURCE_ID});
}
if(!defined($syn_xref_id)){
print STDERR $xref->{ACCESSION}."\n$syn\n";
}
$syn_sth->execute($xref_id, $syn_xref_id ) || die "$dbi->errstr \n $xref_id\n $syn_xref_id\n";
} # foreach syn
......@@ -370,9 +376,9 @@ sub upload_xref_object_graphs {
print STDERR "dbi\t$dbi->err \n$dep{ACCESSION} \n $dep{SOURCE_ID} \n";
}
if(!defined($dep_xref_id)){
print STDERR "$dep{ACCESSION} \n $dep{SOURCE_ID} \n".$dbi->err."\n";
print STDERR "acc = $dep{ACCESSION} \nlink = $dep{LINKAGE_SOURCE_ID} \n".$dbi->err."\n";
}
$dep_sth->execute($xref_id, $dep_xref_id, $dep{LINKAGE_ANNOTATION}, $dep{SOURCE_ID} ) || die $dbi->errstr;
$dep_sth->execute($xref_id, $dep_xref_id, $dep{LINKAGE_ANNOTATION}, $dep{LINKAGE_SOURCE_ID} ) || die $dbi->errstr;
# TODO linkage anntation?
} # foreach dep
......
misc-scripts/xref_mapping/XrefParser/RefSeqGPFFParser.pm
+ 6
− 1
View file @ b0b66b4d
......@@ -141,24 +141,28 @@ sub create_xrefs {
foreach my $ll (@LocusIDline) {
my %dep;
$dep{SOURCE_ID} = $dependent_sources{LocusLink};
$dep{LINKAGE_SOURCE_ID} = $source_id;
$dep{ACCESSION} = $ll;
push @{$xref->{DEPENDENT_XREFS}}, \%dep;
}
foreach my $mim (@mimline) {
my %dep;
$dep{SOURCE_ID} = $dependent_sources{MIM};
$dep{LINKAGE_SOURCE_ID} = $source_id;
$dep{ACCESSION} = $mim;
push @{$xref->{DEPENDENT_XREFS}}, \%dep;
}
foreach my $med (@medline) {
my %dep;
$dep{SOURCE_ID} = $dependent_sources{MEDLINE};
$dep{LINKAGE_SOURCE_ID} = $source_id;
$dep{ACCESSION} = $med;
push @{$xref->{DEPENDENT_XREFS}}, \%dep;
}
foreach my $pub (@pubmed) {
my %dep;
$dep{SOURCE_ID} = $dependent_sources{PUBMED};
$dep{LINKAGE_SOURCE_ID} = $source_id;
$dep{ACCESSION} = $pub;
push @{$xref->{DEPENDENT_XREFS}}, \%dep;
}
......@@ -169,6 +173,7 @@ sub create_xrefs {
if($mrna){
my %mrna_dep;
$mrna_dep{SOURCE_ID} = $source_id; # source is still RefSeq
$mrna_dep{LINKAGE_SOURCE_ID} = $source_id;
my ($mrna_acc,$mrna_ver) = split (/\./,$mrna);
$mrna_dep{ACCESSION} = $mrna_acc;
......@@ -185,7 +190,7 @@ sub create_xrefs {
print "Read " . scalar(@xrefs) ." xrefs from $file\n";
return @xrefs;
return \@xrefs;
}
......
misc-scripts/xref_mapping/XrefParser/RefSeqParser.pm
+ 1
− 1
View file @ b0b66b4d
......@@ -121,7 +121,7 @@ sub create_xrefs {
print "Read " . scalar(@xrefs) ." xrefs from $file\n";
return @xrefs;
return \@xrefs;
}
......
misc-scripts/xref_mapping/XrefParser/UniProtParser.pm
+ 5
− 1
View file @ b0b66b4d
......@@ -195,6 +195,7 @@ sub create_xrefs {
# create dependent xref structure & store it
my %dep;
$dep{SOURCE_NAME} = $source;
$dep{LINKAGE_SOURCE_ID} = $xref->{SOURCE_ID};
$dep{SOURCE_ID} = $dependent_sources{$source};
$dep{ACCESSION} = $acc;
push @{$xref->{DEPENDENT_XREFS}}, \%dep; # array of hashrefs
......@@ -205,6 +206,7 @@ sub create_xrefs {
my %dep2;
$dep2{SOURCE_NAME} = $source;
$dep2{SOURCE_ID} = $dependent_sources{protein_id};
$dep2{LINKAGE_SOURCE_ID} = $xref->{SOURCE_ID};
$dep2{ACCESSION} = $protein_id;
push @{$xref->{DEPENDENT_XREFS}}, \%dep2; # array of hashrefs
}
......@@ -219,6 +221,7 @@ sub create_xrefs {
my %medline_dep;
$medline_dep{SOURCE_ID} = $dependent_sources{PUBMED};
$medline_dep{LINKAGE_SOURCE_ID} = $xref->{SOURCE_ID};
$medline_dep{ACCESSION} = $medline;
push @{$xref->{DEPENDENT_XREFS}}, \%medline_dep;
......@@ -229,6 +232,7 @@ sub create_xrefs {
my %pubmed_dep;
$pubmed_dep{SOURCE_ID} = $dependent_sources{PUBMED};
$pubmed_dep{LINKAGE_SOURCE_ID} = $xref->{SOURCE_ID};
$pubmed_dep{ACCESSION} = $pubmed;
push @{$xref->{DEPENDENT_XREFS}}, \%pubmed_dep;
......@@ -242,7 +246,7 @@ sub create_xrefs {
print "Read $num_sp SwissProt xrefs and $num_sptr SPTrEMBL xrefs from $file\n";
return @xrefs;
return \@xrefs;
#TODO - currently include records from other species - filter on OX line??
}
......
misc-scripts/xref_mapping/sql/table.sql
+ 1
− 3
View file @ b0b66b4d
......@@ -28,7 +28,6 @@ CREATE TABLE primary_xref (
sequence mediumtext,
sequence_type enum('dna','peptide'),
status enum('experimental','predicted'),
source_id int unsigned not null,
PRIMARY KEY (xref_id)
......@@ -41,7 +40,7 @@ CREATE TABLE dependent_xref (
master_xref_id int unsigned not null,
dependent_xref_id int unsigned not null,
linkage_annotation varchar(255),
source_id int unsigned not null,
linkage_source_id int unsigned not null,
KEY master_idx(master_xref_id),
KEY dependent_idx(dependent_xref_id)
......@@ -54,7 +53,6 @@ CREATE TABLE synonym (
xref_id int unsigned not null,
synonym_xref_id int unsigned not null,
source_id int unsigned not null,
KEY xref_idx(xref_id)
......
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment