Accommodated change of SWISSPROT and SPTREMBL to Uniprot/SWISSPROT and

Uniprot/SPTREMBL and added support for tetraodon. For what it's worth.

misc-scripts/protein_match/exonerate_wrapper.pl

@@ -220,12 +220,14 @@ $s_min /= 100.0;
 my %r;
 
 while (defined(my $line = <$in>)) {
+
     # Perl script to calculate the percentage of identity
     # and reformat cigar lines.  Takes tab-delimited list in
     # specific format from exonerate as input on stdin and
     # writes comma-separated output to stdout.
 
-    next if (($line =~ /^Message:/) || ($line =~ /^--/));
+    next if $line !~ /\S+/;
+    next if (($line =~ /^Message:/) || ($line =~ /^Command line:/) || ($line =~ /^--/));
     chomp($line);
 
     # Pick out the individual fields (variable names correspond to

misc-scripts/protein_match/get_Xmapping.pl

@@ -201,6 +201,21 @@ if ($organism eq "fugu") {
 }
 
 
+if ($organism eq "tetraodon") {
+    $check{'sptr_swiss'} = $conf{'sptr_swiss'};
+    $check{'x_map_out'} = $conf{'x_map_out'};
+
+    foreach my $k (keys %check) {
+	print STDERR $check{$k}."\n";
+	if ($check{$k} !~ /(\S+)/) {
+	    usage();
+	}
+    }
+
+}
+
+
+
 #End of check
 
 if ((!defined $organism) || (!defined $sptr_swiss) || (!defined $out)) {
@@ -630,9 +645,9 @@ sub process_parsed_sp{
       die "you have no data tag can't decide if data is from swissprot or trembl for id ".$id." ac ".$ac." entry ".$entry."\n"
     }
     if($tag =~ /STANDARD/){
-      $db = 'SWISSPROT';
+      $db = 'Uniprot/SWISSPROT';
     }elsif($tag =~ /PRELIMINARY/){
-      $db = 'SPTREMBL';
+      $db = 'Uniprot/SPTREMBL';
     }elsif ($tag =~ /Prediction-SPTREMBL/){
 	$db = 'prediction-SPTREMBL';
     }else {

misc-scripts/protein_match/load_transcript_display_id.pl

@@ -27,7 +27,7 @@ my $dbname     = $conf{'db'};
 my $host       = $conf{'host'};
 my $user       = $conf{'dbuser'};
 my $pass       = $conf{'password'};
-my $port = $conf{'port'};
+my $port       = $conf{'port'};
 my $organism   = $conf{'organism'};
 my %priority;
 
@@ -37,9 +37,12 @@ $priority{'MarkerSymbol'} = 1000;
 $priority{'wormbase_transcript'} = 1000;
 $priority{'flybase_symbol'} = 1000;
 $priority{'Anopheles_symbol'} = 1000;
-$priority{'SWISSPROT'} = 900;
+$priority{'Genoscope_annotated_gene'} = 1000;
+$priority{'Genoscope_predicted_transcript'} = 1000;
+$priority{'Genoscope_predicted_gene'} = 999;
+$priority{'Uniprot/SWISSPROT'} = 900;
 $priority{'RefSeq'} = 800;
-$priority{'SPTREMBL'} = 700;
+$priority{'Uniprot/SPTREMBL'} = 700;
 $priority{'LocusLink'} = 100;
 
 
@@ -64,6 +67,7 @@ my $transadaptor = $db->get_TranscriptAdaptor();
 my $geneadaptor  = $db->get_GeneAdaptor();
 my $xrefadaptor  = $db->get_DBEntryAdaptor();
 
+if (0) {
 my $query = "select transcript_id from transcript";
 my $sth = $db->prepare($query);
 $sth->execute();
@@ -87,38 +91,14 @@ while(my $id = $sth->fetchrow) {
     $trans->display_xref($display);
     $transadaptor->update($trans);
 }
-
+}
 print STDERR "Done\n";
 print STDERR "Getting gene display_xref_id\n";
 
 
-if ($organism ne "elegans") {
-    my $query1 = "select gene_id from gene";
-    my $sth1 = $db->prepare($query1);
-    $sth1->execute();
-    while(my $gene_id = $sth1->fetchrow) {
-	my $gene = $geneadaptor->fetch_by_dbID($gene_id);
-	my $transcripts = $gene->get_all_Transcripts();
-	my $display;
-	my $current;
-	TRANS:foreach my $trans(@$transcripts) {
-	    my $xref = $transadaptor->get_display_xref($trans);
-	    if(!$xref){
-	      next TRANS;
-	    }
-	    if ($priority{$xref->database} > $current) {
-	      $display = $xref;
-	      $current = $priority{$xref->database};
-	    }
-	}
-      
-	$gene->display_xref($display);
-	$geneadaptor->update($gene);
-       
-    }
-}
+
 #Not sure id it is really needed if wormbase_gene is put in the priority list... Laura?
-elsif ($organism eq "elegans") {
+if ($organism eq "elegans") {
     my $query1 = "select g.gene_id, x.xref_id from gene_stable_id g, xref x, external_db e where g.stable_id = x.display_label and x.external_db_id = e.external_db_id and e.db_name = 'wormbase_gene'";
     my $sth1 = $db->prepare($query1);
     $sth1->execute();
@@ -129,8 +109,8 @@ elsif ($organism eq "elegans") {
 	$gene->display_xref($xref_obj);
 	$geneadaptor->update($gene);
      }
-    my $query1 = "select g.gene_id, x.xref_id from gene_stable_id g, xref x, external_db e where g.stable_id = x.display_label and x.external_db_id = e.external_db_id and e.db_name = 'wormbase_pseudogene'";
-    my $sth1 = $db->prepare($query1);
+    $query1 = "select g.gene_id, x.xref_id from gene_stable_id g, xref x, external_db e where g.stable_id = x.display_label and x.external_db_id = e.external_db_id and e.db_name = 'wormbase_pseudogene'";
+    $sth1 = $db->prepare($query1);
     $sth1->execute();
     
     while(my ($gene_id,$xref)  = $sth1->fetchrow) {
@@ -140,8 +120,8 @@ elsif ($organism eq "elegans") {
 	$geneadaptor->update($gene);
      }
 
-     my $query1 = "select t.transcript_id, x.xref_id from transcript_stable_id t, xref x, external_db e where t.stable_id = x.display_label and x.external_db_id = e.external_db_id and e.db_name = 'wormbase_pseudogene'";
-    my $sth1 = $db->prepare($query1);
+     $query1 = "select t.transcript_id, x.xref_id from transcript_stable_id t, xref x, external_db e where t.stable_id = x.display_label and x.external_db_id = e.external_db_id and e.db_name = 'wormbase_pseudogene'";
+    $sth1 = $db->prepare($query1);
     $sth1->execute();
     
     while(my ($gene_id,$xref)  = $sth1->fetchrow) {
@@ -152,6 +132,55 @@ elsif ($organism eq "elegans") {
      }
     
 }
+elsif ($organism eq "tetraodon") {
+  # switch the priorityies so that we now favour gene xrefs
+  $priority{'Genoscope_predicted_transcript'} = 999;
+  $priority{'Genoscope_predicted_gene'} = 1000;
+  
+  my $query1 = "select gene_id, type from gene";
+  my $sth1 = $db->prepare($query1);
+  $sth1->execute();
+  while (my ($gene_id, $gene_type) = $sth1->fetchrow) {
+    my $gene = $geneadaptor->fetch_by_dbID($gene_id);    
+    my $xrefs = $gene->get_all_DBLinks;
+
+    my $best_xref;
+    foreach my $xref (@$xrefs) {
+      if (not $best_xref or 
+          $priority{$xref->database} > $priority{$best_xref->database}) {
+        $best_xref = $xref;
+      }
+    }
+    
+    $gene->display_xref($best_xref);
+    $geneadaptor->update($gene);
+  }
+}
+else {
+    my $query1 = "select gene_id from gene";
+    my $sth1 = $db->prepare($query1);
+    $sth1->execute();
+    while(my $gene_id = $sth1->fetchrow) {
+	my $gene = $geneadaptor->fetch_by_dbID($gene_id);
+	my $transcripts = $gene->get_all_Transcripts();
+	my $display;
+	my $current;
+	TRANS:foreach my $trans(@$transcripts) {
+	    my $xref = $transadaptor->get_display_xref($trans);
+	    if(!$xref){
+	      next TRANS;
+	    }
+	    if ($priority{$xref->database} > $current) {
+	      $display = $xref;
+	      $current = $priority{$xref->database};
+	    }
+	}
+      
+	$gene->display_xref($display);
+	$geneadaptor->update($gene);
+       
+    }
+}
 
 print STDERR "Done\n";
 

misc-scripts/protein_match/maps2db.pl

@@ -66,7 +66,7 @@ my %embl2sp;
 my %errorflag;
 my %ref_map_pred;
 
-my $help = $conf{'help'};
+$help = $conf{'help'};
 
 print STDERR "$help\n";
 
@@ -419,7 +419,7 @@ MAPPING: while (<MAP>) {
 	    
 	    
 
-	    if (($a->xDB eq "SPTREMBL") || ($a->xDB eq "SWISSPROT") || ($a->xDB eq "RefSeq") || ($a->xDB eq "ANOSUB") || $a->xDB eq 'BRIGGSAE_HYBRID') {
+	    if (($a->xDB eq "Uniprot/SPTREMBL") || ($a->xDB eq "Uniprot/SWISSPROT") || ($a->xDB eq "RefSeq") || ($a->xDB eq "ANOSUB") || $a->xDB eq 'BRIGGSAE_HYBRID') {
 
 		my $dbentry = Bio::EnsEMBL::IdentityXref->new
 		    ( -adaptor => $adaptor,
@@ -431,7 +431,7 @@ MAPPING: while (<MAP>) {
 
 		$dbentry->status($a->stat);
 
-		if (($check eq "yes") && (($a->xDB eq "SPTREMBL") || ($a->xDB eq "SWISSPROT"))) {
+		if (($check eq "yes") && (($a->xDB eq "Uniprot/SPTREMBL") || ($a->xDB eq "Uniprot/SWISSPROT"))) {
 
 		    if (($sp2embl{$a->xAC}) && ($ens2embl{$targetid})) {
 
@@ -521,7 +521,7 @@ MAPPING: while (<MAP>) {
 		    
 	    }
 	}
-    }
+      }
     
 	
     else  {
@@ -740,6 +740,65 @@ if ($organism eq "elegans") {
 }
 
 
+
+if ($organism eq "tetraodon") {
+  my $adaptor = $db->get_DBEntryAdaptor();
+  
+  my $query = ("select tr.translation_id, ts.stable_id, gs.stable_id, ".
+               "g.type from transcript t, gene_stable_id gs, ".
+               "transcript_stable_id ts, gene g, translation tr ".
+               "where t.gene_id = gs.gene_id and ".
+               "t.transcript_id = ts.transcript_id and ".
+               "t.transcript_id = tr.transcript_id and ".
+               "t.gene_id = g.gene_id");
+  my $sth = $db->prepare($query);
+  $sth->execute();
+  while (my @res = $sth->fetchrow) {
+    my $transl_dbid = $res[0];
+    my $transc_stable_id = $res[1];
+    my $gene_stable_id = $res[2];
+    my $gene_type = $res[3];
+
+    my @xrefs = {
+        primary_id => $gene_stable_id, 
+        display_id => $gene_stable_id,
+    };
+
+    if ($gene_type =~ /annotated/i) {
+      $xrefs[0]->{dbname} = "Genoscope_annotated_gene";
+      $xrefs[0]->{status} = "KNOWN";
+    } else {
+      $xrefs[0]->{dbname} = "Genoscope_predicted_gene";
+      $xrefs[0]->{status} = "XREF";
+      push @xrefs, {
+        dbname => "Genoscope_predicted_transcript",
+        primary_id => $transc_stable_id, 
+        display_id => $transc_stable_id,
+        status     => "XREF"
+      }
+    }
+
+    foreach my $xref (@xrefs) {
+
+      my $dbentry = Bio::EnsEMBL::DBEntry->new
+          ( -adaptor => $adaptor,
+            -primary_id => $xref->{primary_id},
+            -display_id => $xref->{display_id},
+            -version => 1,
+            -release => 1,
+            -dbname => $xref->{dbname});
+
+      $dbentry->status($xref->{status});            
+      $adaptor->store($dbentry,$transl_dbid,"Translation");
+    }
+  }
+}
+
+
+
+
+
+
 sub usage {
     
     print STDERR <<HELP