Changeset 517

Timestamp:

05/26/11 04:28:12 (12 months ago)

Author:

AlexanderPico

Message:

adapted to collect probes and probesets, skipping problematic probe_features and Ex- (all exon) arrays

Files:

• trunk/dbbuilder/src/org/bridgedb/extract/Ensembl_ETL_Device_v18_local.pl (modified) (5 diffs)

trunk/dbbuilder/src/org/bridgedb/extract/Ensembl_ETL_Device_v18_local.pl

@@ -427 +427 @@
 my $slice_adaptor = $registry->get_adaptor($species, "core", "slice");
 my $go_adaptor = $registry->get_adaptor("Multi", "Ontology", "GOTerm");
-my $probe_adaptor = $registry->get_adaptor($species, "funcgen", "ProbeFeature");
+my $probe_feature_adaptor = $registry->get_adaptor($species, "funcgen", "ProbeFeature");
+my $probe_adaptor = $registry->get_adaptor($species, "funcgen", "Probe");
+my $probe_set_adaptor = $registry->get_adaptor($species, "funcgen", "Probeset");
 my @dbas = @{Bio::EnsEMBL::Registry->get_all_DBAdaptors(-species => $species)};
 my $dbname = $dbas[0]->dbc->dbname();        # e.g., core_mus_musculus_42_36c
@@ -2178 +2180 @@
   ## Solution #2 only works after API method has been added
   ## See NathJohnson_API_patch.txt
-    my $probe_features = $probe_adaptor->fetch_all_by_linked_transcript_Gene($gene);
-  
-    foreach my $pf (@$probe_features) {
-      my $probe = $pf->probe();
+    #my @probe_features = @{$probe_feature_adaptor->fetch_all_by_linked_transcript_Gene($gene)};
+    my @probes = @{$probe_adaptor->fetch_all_by_linked_transcript_Gene($gene)};
+    my @probe_sets = @{$probe_set_adaptor->fetch_all_by_linked_transcript_Gene($gene)};
+    my @all_probes = (@probes, @probe_sets);
+
+    foreach my $probe (@all_probes) {
+      #my $probe = $pf->probe();
       my $array_list = $probe->get_all_Arrays();
 
@@ -2189 +2194 @@
         }
         my $pf_dbname = mysql_quotes($array->vendor());
-        my $pf_display_id = mysql_quotes($probe->get_probename($array->name()));
+        my $pf_display_id = 'null';
         if ($pf_dbname =~ /^\'AFFY/i) { # Affy uses probesets
-                $pf_display_id = mysql_quotes($probe->probeset()->name());
-        }
+                $pf_display_id = mysql_quotes($probe->name()); ##probeset()->name());
+        } else {
+                $pf_display_id = mysql_quotes($probe->get_probename($array->name()));
+        }
+                
         my $pf_primary_id = $pf_display_id;
         my $pf_description = mysql_quotes($array->description());
@@ -2200 +2208 @@
         my $pf_info_text = mysql_quotes($array->type());
         my $pf_info_type = mysql_quotes("SEQUENCE_MATCH"); #enum
-        my $pf_synonyms = $probe->get_all_probenames();
-        my $pf_syns = join("|", @$pf_synonyms);
+        #my $pf_synonyms = $probe->get_all_probenames();
+        my $pf_syns = ''; ##join("|", @$pf_synonyms);
         $pf_syns = mysql_quotes($pf_syns);
 
@@ -2209 +2217 @@
                              $pf_release, $pf_status, $pf_version, $pf_info_text, $pf_info_type];
 
-        if ($pf_dbname =~ /^\'AFFY/i){  #catch all types
+        if ($pf_dbname =~ /^\'AFFY/i && $pf_release !~ /Ex-/){  #catch all types #skip all exon arrays
             $ADMIN_Xrefs{$pf_dbname}[10] = "\'Y\'"; # collected
             if (!${$seen{Affy}{$pf_primary_id}}++){