hack fix for refseq transcripts missing gene symbol data

Ensembl · Mar 31, 2015 · e06a7d8 · e06a7d8
1 parent df29344
commit e06a7d8
Showing 1 changed file with 13 additions and 3 deletions.
diff --git a/modules/Bio/EnsEMBL/Variation/Utils/VEP.pm b/modules/Bio/EnsEMBL/Variation/Utils/VEP.pm
@@ -3445,6 +3445,7 @@ sub fetch_transcripts {
 
             ## hack to copy HGNC IDs
             my %hgnc_ids = ();
+            my %refseq_stuff = ();
 
             # add loaded transcripts to main cache
             if(defined($tmp_cache->{$chr})) {
@@ -3500,14 +3501,23 @@ sub fetch_transcripts {
                     ## hack to copy HGNC IDs
                     $hgnc_ids{$tr->{_gene_symbol}} = $tr->{_gene_hgnc_id} if defined($tr->{_gene_hgnc_id});
 
+                    ## hack to copy RefSeq gene stuff
+                    if(defined($config->{refseq}) || defined($config->{merged})) {
+                      $refseq_stuff{$tr->{_gene}->stable_id}->{$_} ||= $tr->{$_} for qw(_gene_symbol _gene_symbol_source _gene_hgnc_id);
+                    }
+
                     $seen_trs{$dbID} = 1;
 
                     push @{$tr_cache->{$chr}}, $tr;
                 }
 
-                ## hack to copy HGNC IDs
-                for(@{$tr_cache->{$chr}}) {
-                  $_->{_gene_hgnc_id} = $hgnc_ids{$_->{_gene_symbol}} if defined($_->{_gene_symbol}) && defined($hgnc_ids{$_->{_gene_symbol}});
+                ## hack to copy HGNC IDs and RefSeq stuff
+                foreach my $tr(@{$tr_cache->{$chr}}) {
+                  $tr->{_gene_hgnc_id} = $hgnc_ids{$tr->{_gene_symbol}} if defined($tr->{_gene_symbol}) && defined($hgnc_ids{$tr->{_gene_symbol}});
+
+                  if(defined($config->{refseq}) || defined($config->{merged})) {
+                    $tr->{$_} ||= $refseq_stuff{$tr->{_gene}->stable_id}->{$_} for qw(_gene_symbol _gene_symbol_source _gene_hgnc_id);
+                  }
                 }
             }