SwissProt records with multiple accessions now create appropriate records in...

SwissProt records with multiple accessions now create appropriate records in the xref and synonym tables

misc-scripts/xref_mapping/BaseParser.pm

@@ -58,17 +58,31 @@ sub upload_xrefs {
 		  $xref->{SPECIES_ID}) || die $dbi->errstr;
 
     # get ID of xref just inserted
-     my $id = $sth->{'mysql_insertid'};
+    my $xref_id = $sth->{'mysql_insertid'};
 
     # create entry in primary_xref table with sequence
     # TODO experimental/predicted????
     $sth = $dbi->prepare("INSERT INTO primary_xref VALUES(?,?,?,?,?)");
-    $sth->execute($id,
+    $sth->execute($xref_id,
 		  $xref->{SEQUENCE},
 		  'peptide',
 		  'experimental',
 		  $xref->{SOURCE_ID}) || die $dbi->errstr;
 
+    # if there are synonyms, create xrefs for them and entries in the synonym table
+    my $xref_sth = $dbi->prepare("INSERT INTO xref (accession,label,source_id,species_id) VALUES(?,?,?,?)");
+    my $syn_sth = $dbi->prepare("INSERT INTO synonym VALUES(?,?,?)");
+    foreach my $syn (@{$xref->{SYNONYMS}}) {
+      
+      $xref_sth->execute($syn,
+			 $xref->{LABEL},
+			 $xref->{SOURCE_ID},
+			 $xref->{SPECIES_ID}) || die $dbi->errstr;
+
+      my $syn_xref_id = $xref_sth->{'mysql_insertid'};
+      $syn_sth->execute($xref_id, $syn_xref_id, $xref->{SOURCE_ID} ) || die $dbi->errstr;
+
+    }
   }
 
   $sth->finish() if defined $sth;

misc-scripts/xref_mapping/SwissProtParser.pm

@@ -107,11 +107,20 @@ sub create_xrefs {
   while (<SWISSPROT>) {
 
     my $xref;
-    ($xref->{ACCESSION}) =$_ =~ /AC\s+(\w+);/; # note only takes first accession if > 1
+    my $acc;
+    ($acc) =$_ =~ /AC\s+(.+);/; # may catch multiple ; separated accessions 
     ($xref->{LABEL})    = $_ =~ /ID\s+(\w+)/;
     ($xref->{SPECIES_ID}) = $species_id;
     ($xref->{SOURCE_ID}) = $source_id;
 
+    # set accession (and synonyms if more than one)
+    # note synonyms 
+    my @acc = split /;/, $acc;
+    $xref->{ACCESSION} = $acc[0];
+    for (my $a=1; $a <= $#acc; $a++) {
+      push(@{$xref->{"SYNONYMS"} }, $acc[$a]);
+    }
+
     # extract sequence
     my ($seq) = $_ =~ /SQ\s+(.+)/s; # /s allows . to match newline 
       my @seq_lines = split /\n/, $seq;