CCDS -> Ensembl transcript is not a one to one mapping so cannot use a key...

CCDS -> Ensembl transcript is not a one to one mapping so cannot use a key value pairs in hash. Hence now the hash uses a anonymous array to stire multiple values

misc-scripts/xref_mapping/XrefParser/HGNC_CCDSParser.pm

@@ -90,7 +90,7 @@ sub run_script {
   $sth->execute() or croak( $dbi2->errstr() );
   while ( my @row = $sth->fetchrow_array() ) {
     if(defined($trans_id_to_stable_id{$row[0]})){
-      $ccds_to_stable_id{$row[1]} = $trans_id_to_stable_id{$row[0]};
+      push @{$ccds_to_stable_id{$row[1]}}, $trans_id_to_stable_id{$row[0]};
     }
     else{
       print "NO transcript_stable_id for  for ".$row[0]."\n";
@@ -145,13 +145,19 @@ sub run_script {
  
   my $xref_count = 0;
   my $no_ccds_to_hgnc = 0;
+  my $direct_count = 0;
   foreach my $ccds (keys %ccds_to_stable_id){
     if(defined($ccds_to_hgnc{$ccds})){
       my $hgnc = $ccds_to_hgnc{$ccds};
       $hgnc =~ s/HGNC://;
       my $xref_id = $self->add_xref($hgnc, $version{$hgnc} , $label{$hgnc}||$hgnc , 
 				      $description{$hgnc}, $source_id, $species_id, "DIRECT");
-      $self->add_direct_xref($xref_id, $ccds_to_stable_id{$ccds}, "Transcript", "");
+
+      foreach my $stable_id (@{$ccds_to_stable_id{$ccds}}){
+	$self->add_direct_xref($xref_id, $stable_id, "Transcript", "");
+	$direct_count++;
+      }	
+
       $xref_count++;
 
       if(defined($syn_hash->{$hgnc})){
@@ -168,7 +174,7 @@ sub run_script {
     }
   }
   $add_syn_sth->finish;
-  print "$no_ccds_to_hgnc missed as no hgnc for the ccds. Added $xref_count HGNC xrefs via CCDS\n" if($verbose);
+  print "$no_ccds_to_hgnc missed as no hgnc for the ccds. Added $xref_count HGNC xrefs via CCDS and $direct_count direct xrefs\n" if($verbose);
   return 0;
 }
 

misc-scripts/xref_mapping/XrefParser/RefSeq_CCDSParser.pm

@@ -73,7 +73,7 @@ sub run_script {
   my %ensembl_type;
   my %old_xref;
   while (my @row = $sth->fetchrow_array()) {
-      $ensembl_stable_id{$access} = $stable_id;
+      push @{$ensembl_stable_id{$access}}, $stable_id;
       $ensembl_type{$access} = $type;
       $old_xref{$access} = $old_xref_id; 
   }
@@ -83,9 +83,13 @@ sub run_script {
 
   my $line_count = 0;
   my $xref_count = 0;
+  my $direct_count = 0;
   my %seen;
   my %old_to_new;
 
+#
+# dbi2 is the ccds database
+#
   my $dbi2 = $self->dbi2($host, $port, $user, $dbname, $pass);
   if(!defined($dbi2)){
     return 1;
@@ -97,8 +101,9 @@ sub run_script {
   # get ccds -> xref transcript_id                 ensembl_stable_id{CCDS1} = ENST00001
   # get ccds -> internal transcript_id             ccds_to_internal_id(CCDS1} = 12345
 
-  $sql = 'select x.dbprimary_acc, ox.ensembl_id from xref x, object_xref ox, external_db e where x.xref_id = ox.xref_id and x.external_db_id = e.external_db_id and e.db_name like ?';
+  $sql = 'select x.dbprimary_acc, ox.ensembl_id from xref x, object_xref ox, external_db e where x.xref_id = ox.xref_id and x.external_db_id = e.external_db_id and e.db_name like ? order by x.version';
 
+# order by version added so that the hash gets overwritten with the latest version.
 
 
   # calculate internal_id -> xref transcript_id
@@ -141,7 +146,11 @@ sub run_script {
       my $xref_id = $self->add_xref($refseq, $version{$refseq} , $label{$refseq}||$refseq , 
 				    $description{$refseq}, $new_source_id, $species_id, "DIRECT");
 
-      $self->add_direct_xref($xref_id, $internal_to_stable_id{$internal_id}, "Transcript", "");
+
+      foreach my $stable_id (@{$internal_to_stable_id{$internal_id}}){
+	$self->add_direct_xref($xref_id, $stable_id, "Transcript", "");
+	$direct_count++;
+      }
 
       $old_to_new{$old_xref{$refseq}} = $xref_id;
       $xref_count++;
@@ -166,7 +175,7 @@ sub run_script {
   }
 
 
-  print "Parsed $line_count RefSeq_dna identifiers from $file, added $xref_count xrefs and $xref_count direct_xrefs  from $line_count lines.\n" if ($verbose);
+  print "Parsed $line_count RefSeq_dna identifiers from $file, added $xref_count xrefs and $direct_count direct_xrefs  from $line_count lines.\n" if ($verbose);
 
 
   return 0;