look up stable_ids in core and create direct xrefs

misc-scripts/xref_mapping/XrefParser/ArrayExpressParser.pm

@@ -13,51 +13,133 @@ use strict;
 use warnings;
 use Carp;
 use base qw( XrefParser::BaseParser );
+use Bio::EnsEMBL::Registry;
 
-sub meta_key {
-  my ($self) = @_;
-  return "array_express.exported";
-}
 
-sub run {
+sub run_script {
+
   my ($self, $ref_arg) = @_;
-  
   my $source_id    = $ref_arg->{source_id};
   my $species_id   = $ref_arg->{species_id};
-  my $files        = $ref_arg->{files};
+  my $file         = $ref_arg->{file};
   my $verbose      = $ref_arg->{verbose};
-  
-  if((!defined $source_id) or (!defined $species_id) or (!defined $files) ){
-    croak "Need to pass source_id, species_id and files as pairs";
+
+  if((!defined $source_id) or (!defined $species_id) or (!defined $file) ){
+    croak "Need to pass source_id, species_id and file as pairs";
+  }
+  $verbose |=0;
+
+  my $project;
+  my $wget = "";
+
+  if($file =~ /project[=][>](\S+?)[,]/){
+    $project = $1;
+  }
+  if($file =~ /wget[=][>](\S+?)[,]/){
+    $wget = $1;
+  }
+
+
+  my $ua = LWP::UserAgent->new();
+  $ua->timeout(10);
+  $ua->env_proxy();
+
+  my $response = $ua->get($wget);
+
+  if ( !$response->is_success() ) {
+    warn($response->status_line);
+    return 1;
   }
-  $verbose ||=0;
+  my @lines = split(/\n/,$response->content);
 
   my %species_id_to_names = $self->species_id2name();
   my $species_id_to_names = \%species_id_to_names;
   my $names = $species_id_to_names->{$species_id};
-  my $contents_lookup = $self->_get_contents($files, $verbose);
+  my $contents_lookup = $self->_get_contents(\@lines, $verbose);
   my $active = $self->_is_active($contents_lookup, $names, $verbose);
+ 
+  if (!$active) {
+      return;
+  }
 
-  if ($active && $verbose) {
-      print "ArrayExpress xrefs will be created when running xref_mapper.pl/DirectXrefs.pm as gene stable ids are required to create the xrefs\n";
+  #get stable_ids from core and create xrefs 
+
+  my $registry = "Bio::EnsEMBL::Registry";
+
+  if ($project eq 'ensembl') {
+      $registry->load_registry_from_multiple_dbs( 
+	  {
+	      '-host'    => 'ens-staging1',
+	      '-user'    => 'ensro',
+	  },
+	  {
+	      '-host'     => 'ens-staging2',
+	      '-user'     => 'ensro',
+	  },
+       );
+  } elsif ($project eq 'ensemblgenomes') {
+
+      $registry->load_registry_from_multiple_dbs( 
+	  {
+	      '-host'     => 'mysql-eg-staging-1.ebi.ac.uk',
+	      '-port'     => 4160,
+	      '-user'     => 'ensro',
+	  },
+	  {
+	      '-host'     => 'mysql-eg-staging-2.ebi.ac.uk',
+	      '-port'     => 4275,
+	      '-user'     => 'ensro',
+	  },
+ 
+      );
+
+  } else {
+      die("Missing or unsupported project value. Supported values: ensembl, ensemblgenomes");
   }
-  $self->add_meta_pair($self->meta_key(),$active);
-  $self->add_meta_pair('species_id',$species_id);
 
-  return;
+  #get the species name
+ 
+  my $species_name = $species_id_to_names{$species_id}[0];
+  my $gene_adaptor = $registry->get_adaptor($species_name, 'core', 'Gene');
+
+  my @stable_ids = map { $_->stable_id } @{$gene_adaptor->fetch_all()};
+
+  my $xref_count = 0;
+  foreach my $gene_stable_id (@stable_ids) {
+      
+      my $xref_id = $self->add_xref({ acc        => $gene_stable_id,
+					 label      => $gene_stable_id,
+					 source_id  => $source_id,
+					 species_id => $species_id,
+					 info_type => "DIRECT"} );
+	
+      $self->add_direct_xref( $xref_id, $gene_stable_id, 'gene', '');
+      if ($xref_id) {
+	 $xref_count++;
+      }
+  }
+	   
+  print "Added $xref_count DIRECT xrefs\n" if($verbose);
+  if ( !$xref_count ) {
+      return 1;    # 1 error
+  }
+
+  return 0; # successfull	   
+
 }
 
+
+
 sub _get_contents {
-  my ($self, $files, $verbose) = @_;
+  my ($self, $lines, $verbose) = @_;
+  my @lines = @$lines;
   my %lookup;
-  my $fh = $self->get_filehandle($files->[0]);
-  while(my $line = <$fh>) {
-    chomp $line;
+
+  foreach my $line (@lines) {
     my ($species, $remainder) = $line =~ /^([a-z|A-Z]+)_(.+)$/;
     croak "The line '$line' is not linked to a gene set. This is unexpected." if $remainder !~ /gene/;
     $lookup{$species} = 1;
   }
-  close ($fh);
   if($verbose) {
     printf("ArrayExpress is using the species [%s]\n", join(q{, }, keys %lookup));
   }
@@ -90,53 +172,5 @@ sub _is_active {
   return $active;
 }
 
-#this method is called from XrefMapper/DirectXrefs.pm
-
-sub create_xrefs {
-    my $self = shift;
-    my $verbose = shift;
-
-    my $array_xrefs_meta_key = $self->meta_key();
-  
-    if ($array_xrefs_meta_key) {
-      my $active = $self->get_meta_value($array_xrefs_meta_key);
-      if ($active) {
-	   #create ArrayExpress direct xrefs
-	   my $source_name = 'ArrayExpress';
-	   my $source_id = $self->get_source_id_for_source_name($source_name);
-	   
-	   my $species_id = $self->get_meta_value('species_id');
-	   #get gene stable_ids
-	   my $gene_id_sth = $self->dbi()->prepare("select stable_id from gene_stable_id order by stable_id");
-	   $gene_id_sth->execute();
-	   my $gene_stable_id;
-	   $gene_id_sth->bind_columns(\$gene_stable_id);
-	   my $xref_count = 0;
-	   while ($gene_id_sth->fetch()) {
-	       
-	       my $xref_id = $self->add_xref({ acc        => $gene_stable_id,
-					 label      => $gene_stable_id,
-					 source_id  => $source_id,
-					 species_id => $species_id,
-					 info_type => "DIRECT"} );
-	
-	       $self->add_direct_xref( $xref_id, $gene_stable_id, 'gene', '');
-	       if ($xref_id) {
-		   $xref_count++;
-	       }
-	   }
-	   $gene_id_sth->finish();
-	   
-	   if ($xref_count > 0) {
-	       print "Loaded $xref_count $source_name DIRECT xrefs\n" if $verbose;	       
-	   } else {
-
-	       print "Warning: 0 $source_name DIRECT xrefs loaded even though $source_name is active for the species.\n";
-	   }
-      }
-
-    }
-}
-
 
 1;