anopholese multi tax_id fix

misc-scripts/xref_mapping/XrefParser/BaseParser.pm

@@ -101,7 +101,7 @@ sub run
   }
 
   my $sql =
-    "SELECT s.source_id, su.source_url_id, s.name, su.url, su.checksum, su.parser, su.species_id, sp.name " .
+    "SELECT distinct(s.source_id), su.source_url_id, s.name, su.url, su.checksum, su.parser, su.species_id " .
       "FROM source s, source_url su, species sp " .
 	"WHERE s.download='Y' AND su.source_id=s.source_id " .
           "AND su.species_id=sp.species_id " .
@@ -113,10 +113,10 @@ sub run
   $sth->execute();
 
   my ( $source_id, $source_url_id, $name, $url, $checksum, $parser,
-      $species_id, $species_name );
+      $species_id);
 
   $sth->bind_columns( \$source_id, \$source_url_id, \$name, \$url,
-      \$checksum, \$parser, \$species_id, \$species_name );
+      \$checksum, \$parser, \$species_id);
 
   my $last_type = "";
   my $dir;
@@ -154,7 +154,7 @@ sub run
 	print "Parsing $dsn with $parser\n";
         eval "require XrefParser::$parser";
         my $new = "XrefParser::$parser"->new();
-        if($new->run($dsn, $source_id, $species_id, $name, $species_name)){
+        if($new->run($dsn, $source_id, $species_id, $name, undef)){
 	  $summary{$parser}++;
 	}
 	next;
@@ -1241,16 +1241,18 @@ sub show_valid_species() {
 
 sub get_taxonomy_from_species_id{
   my ($self,$species_id) = @_;
+  my %hash;
 
   my $dbi = dbi();
   my $sth = $dbi->prepare("SELECT taxonomy_id FROM species WHERE species_id = $species_id");
   $sth->execute() or croak( $dbi->errstr() );
-  if(my @row = $sth->fetchrow_array()) {
-    return $row[0];
+  while(my @row = $sth->fetchrow_array()) {
+    $hash{$row[0]} = 1;
   }   
   $sth->finish;
-  return undef;
+  return \%hash;
 }
+
 sub get_direct_xref{
   my ($self,$stable_id,$type,$link) = @_;
 

misc-scripts/xref_mapping/XrefParser/EntrezGeneParser.pm

@@ -34,7 +34,7 @@ sub run {
     $species_id = XrefParser::BaseParser->get_species_id_for_filename($file);
   }
   
-  my $species_tax_id = $self->get_taxonomy_from_species_id($species_id);
+  my %species_tax_id = %{$self->get_taxonomy_from_species_id($species_id)};
   
 
     my $eg_io = $self->get_filehandle($file);
@@ -92,7 +92,7 @@ sub run {
   while ( $_ = $eg_io->getline() ) {
     chomp;
     my (@arr) = split(/\t/,$_);
-    if($arr[$gene_tax_id_index] != $species_tax_id){
+    if(!defined($species_tax_id{$arr[$gene_tax_id_index]})){
       next;
     }
     my $acc    = $arr[$gene_id_index];

misc-scripts/xref_mapping/XrefParser/FastaParser.pm

@@ -17,7 +17,7 @@ sub run {
   my ($self, $source_id, $species_id, $file) = @_;
   
   my $sio = Bio::SeqIO->new(-format=>'fasta' , -file=>$file );
-  my $species_tax_id = $self->get_taxonomy_from_species_id($species_id);
+  my %species_tax_id = %{$self->get_taxonomy_from_species_id($species_id)};
   
   my @xrefs;
   while( my $seq = $sio->next_seq ) {
@@ -25,7 +25,7 @@ sub run {
     # Test species if available
     if( my $sp = $seq->species ){
       if( my $tax_id = $sp->ncbi_taxid ){
-        next if $tax_id != $species_tax_id;
+        next if (!defined $species_tax_id{$tax_id});
       }
     }
 

misc-scripts/xref_mapping/XrefParser/IPIParser.pm

@@ -26,7 +26,7 @@ sub run {
     return 1;    # 1 = error
   }
 
-  my $species_tax_id = $self->get_taxonomy_from_species_id($species_id);
+  my %species_tax_id = %{$self->get_taxonomy_from_species_id($species_id)};
 
   while ( $_ = $ipi_io->getline() ) {
     my $xref;
@@ -43,7 +43,7 @@ sub run {
     # note currently we ignore all the other cross-references in the IPI file
 
     # only interested in species with the taxonomy ID were looking for
-    next if ( !defined $tax_id || $tax_id != $species_tax_id );
+    next if ( !defined $tax_id || !defined $species_tax_id{$tax_id});
 
     # make sequence into one long string
     $sequence =~ s/\n//g;

misc-scripts/xref_mapping/XrefParser/UniProtVarSplicParser.pm

@@ -31,7 +31,6 @@ sub run {
     return 1;    # 1 error
   }
 
-  my $species_tax_id = $self->get_taxonomy_from_species_id($species_id);
   my %swiss = %{ $self->get_valid_codes( "uniprot", $species_id ) };
 
   my $missed = 0;

misc-scripts/xref_mapping/XrefParser/XenopusJamboreeParser.pm

@@ -28,8 +28,6 @@ sub run {
     return 1;    # 1 error
   }
 
-  my $species_tax_id = $self->get_taxonomy_from_species_id($species_id);
-
   while ( $_ = $file_io->getline() ) {
     my $xref;