V2 Illumina parser for mouse

misc-scripts/xref_mapping/XrefParser/IlluminaWGParser.pm

+package XrefParser::IlluminaWGParser;
+
+use strict;
+
+use base qw( XrefParser::BaseParser );
+
+
+sub run {
+
+  my ($self, $source_id, $species_id, $file) = @_;
+
+  my @xrefs;
+
+  my $file_io = $self->get_filehandle($file);
+
+  if ( !defined $file_io ) {
+    print "Could not open $file\n";
+    return 1;
+  }
+
+  my $read = 0;
+  my $name_index;
+  my $seq_index;
+  my $defin_index;
+  while ( $_ = $file_io->getline() ) {
+    chomp;
+
+    my $xref;
+
+    # strip ^M at end of line
+    $_ =~ s/\015//g;
+
+    if(/^\[/){
+      print $_."\n";
+      if(/^\[Probes/){
+	my $header = $file_io->getline();
+	print $header."\n";
+	$read =1;
+	my @bits = split("\t", $header);
+	my $index =0;
+	foreach my $head (@bits){
+	  if($head eq "Search_Key"){
+	    $name_index = $index;
+	  }
+	  elsif($head eq "Probe_Sequence"){
+	    $seq_index = $index;
+	  }
+	  elsif($head eq "Definition"){
+	    $defin_index = $index;
+	  }
+	  $index++;
+	}
+	if(!defined($name_index) or !defined($seq_index) or !defined($defin_index)){
+	  die "Could not find index for search_key->$name_index, seq->$seq_index, definition->$defin_index";
+	}
+      
+	next;
+      }
+      else{
+	$read = 0;
+      }	
+    }
+    if($read){
+#      print $_."\n";
+      my @bits = split("\t", $_);
+      my $sequence = $bits[$seq_index];
+
+      my $description = $bits[$defin_index];
+      my $illumina_id = $bits[$name_index];
+
+      # build the xref object and store it
+      $xref->{ACCESSION}     = $illumina_id;
+      $xref->{LABEL}         = $illumina_id;
+      $xref->{SEQUENCE}      = $sequence;
+      $xref->{SOURCE_ID}     = $source_id;
+      $xref->{SPECIES_ID}    = $species_id;
+      $xref->{DESCRIPTION}   = $description;
+      $xref->{SEQUENCE_TYPE} = 'dna';
+      $xref->{STATUS}        = 'experimental';
+
+      push @xrefs, $xref;
+      
+    }
+  }
+
+  $file_io->close();
+
+  print scalar(@xrefs) . " Illumina V2 xrefs succesfully parsed\n";
+
+  XrefParser::BaseParser->upload_xref_object_graphs(\@xrefs);
+
+
+  return 0;
+}
+
+1;

misc-scripts/xref_mapping/xref_config.ini

@@ -854,6 +854,17 @@ parser          = FastaParser
 release_uri     =
 data_uri        = file:Illumina_V1/Mouse-6_V1.fa
 
+[source Illumina_V2::mus_musculus]
+# Used by mus_musculus
+name            = Illumina_V2
+download        = Y
+order           = 50
+priority        = 1
+prio_descr      =
+parser          = IlluminaWGParser
+release_uri     =
+data_uri        = file:Illumina_V2/MouseWG-6_V2_0_R1_11278593_A.txt
+
 [source Illumina_V2::homo_sapiens]
 # Used by homo_sapiens
 name            = Illumina_V2
@@ -2576,6 +2587,7 @@ source          = InterproGO::MULTI
 source          = IMGT/GENE_DB::mus_musculus
 source          = IPI::mus_musculus
 source          = Illumina_V1::mus_musculus
+source          = Illumina_V2::mus_musculus
 source          = Interpro::MULTI
 source          = OTTT::MULTI
 source          = RefSeq_dna::mus_musculus