diff --git a/misc-scripts/xref_mapping/XrefParser/BaseParser.pm b/misc-scripts/xref_mapping/XrefParser/BaseParser.pm
index 9b9fc46bc8a768697e6bd22904972e6e6d7c9151..145c320ff653290a883ce654efb43df61bf2a0d2 100644
--- a/misc-scripts/xref_mapping/XrefParser/BaseParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/BaseParser.pm
@@ -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
diff --git a/misc-scripts/xref_mapping/XrefParser/RefSeqGPFFParser.pm b/misc-scripts/xref_mapping/XrefParser/RefSeqGPFFParser.pm
index bfc69b08812bedeb265e0745641c9fc5feed96cd..8a677b95416e38c804bd42bdf886efd69085ba30 100644
--- a/misc-scripts/xref_mapping/XrefParser/RefSeqGPFFParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/RefSeqGPFFParser.pm
@@ -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;
}
diff --git a/misc-scripts/xref_mapping/XrefParser/RefSeqParser.pm b/misc-scripts/xref_mapping/XrefParser/RefSeqParser.pm
index 7ea1d3f5a97a6a3a5acc66e325c62bd3f5e64673..83f3eb3f72a5e7496e84dbe59f606860748d22cd 100644
--- a/misc-scripts/xref_mapping/XrefParser/RefSeqParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/RefSeqParser.pm
@@ -121,7 +121,7 @@ sub create_xrefs {
print "Read " . scalar(@xrefs) ." xrefs from $file\n";
- return @xrefs;
+ return \@xrefs;
}
diff --git a/misc-scripts/xref_mapping/XrefParser/UniProtParser.pm b/misc-scripts/xref_mapping/XrefParser/UniProtParser.pm
index 78449c9f271f01315a7567e2d4d601d0911ae324..90e90d296870d510f4c011ce9de9a03e147e5f11 100644
--- a/misc-scripts/xref_mapping/XrefParser/UniProtParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/UniProtParser.pm
@@ -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??
}
diff --git a/misc-scripts/xref_mapping/sql/table.sql b/misc-scripts/xref_mapping/sql/table.sql
index df13b89a24e8532744892254c2f62231a4be79d5..105e7d6a0113a5e16c57dce5e2cc78d2c802fe0f 100755
--- a/misc-scripts/xref_mapping/sql/table.sql
+++ b/misc-scripts/xref_mapping/sql/table.sql
@@ -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)