Merge branch 'postreleasefix/110'

Ensembl · Apr 19, 2023 · fb90395 · fb90395
2 parents 290f78d + 2b183d6
commit fb90395
Show file tree

Hide file tree

Showing 2 changed files with 50 additions and 3 deletions.
diff --git a/modules/Bio/EnsEMBL/Variation/DBSQL/VariationFeatureAdaptor.pm b/modules/Bio/EnsEMBL/Variation/DBSQL/VariationFeatureAdaptor.pm
@@ -2368,12 +2368,48 @@ sub get_reference{
 
   my $tr_mapper = $transcript->get_TranscriptMapper(); 
 
+  # If the codon overlaps an exon-intron boundary @coords can return two coordinates:
+  #  one for the first exon where the codon starts
+  #  a second coordinate for the exon where the codon ends
   my @coords = defined($pos2) ? $tr_mapper->pep2genomic($pos, $pos2) : $tr_mapper->pep2genomic($pos, $pos);  
 
-  my $start  = $coords[0]->start();
-  my $end    = $coords[0]->end();
+  my $start;
+  my $end;
+  my $real_start;
+  my $real_end;
   my $strand = $coords[0]->strand();
 
+  # Assign start and end when we have two coordinates
+  # This is the most common alternative
+  if(scalar(@coords) == 2){
+    if(($coords[0]->end() - $coords[0]->start() + 1) % 2 == 0) {
+      if($strand == 1) {
+        $start = $coords[0]->start();
+        $end = $start + 2;
+        $real_end = $coords[1]->end();
+      }
+      else {
+        $end = $coords[0]->end();
+        $start = $end - 2;
+        $real_start = $coords[1]->start();
+      }
+    }
+    elsif($strand == 1) {
+      $end = $coords[1]->end();
+      $start = $end - 2;
+      $real_start = $coords[0]->start();
+    }
+    else {
+      $start = $coords[0]->start();
+      $end = $start + 2;
+      $real_end = $coords[1]->end();
+    }
+  }
+  else{
+    $start = $coords[0]->start();
+    $end   = $coords[0]->end();
+  }
+
   my $seq_length = $type_del == 1 ? ($end-$start) + 1 : 3;  
 
   ## find reference sequence 
@@ -2390,8 +2426,11 @@ sub get_reference{
 
   my $from_codon_ref = $from_slice->seq(); 
 
+  $start = $real_start ? $real_start : $start;
+  $end = $real_end ? $real_end : $end;
+
   ## correct for strand
-  reverse_comp(\$from_codon_ref) if $strand <0; 
+  reverse_comp(\$from_codon_ref) if $strand <0;
 
   return ($from_codon_ref, $start, $end, $strand); 
 }

diff --git a/modules/t/variationFeatureAdaptor.t b/modules/t/variationFeatureAdaptor.t
@@ -361,6 +361,7 @@ ok($vfs16a->[0]->variation_name() eq $vf_somatic_name, "somatic vf with phenotyp
 # test fetching VF with empty consequence type column
 is($vfa->fetch_by_dbID(997738282)->display_consequence, 'sequence_variant', 'empty consequence column');
 
+ok(scalar @{$vfa->fetch_all_by_location_identifier('18:40228819:A_G')} == 1, "fetch_all_by_location_identifier '18:40228819:A_G'");
 
 # test fetch Iterator
 print "\n# Test - fetch_Iterator\n";
@@ -445,6 +446,13 @@ ok($vfa->fetch_by_hgvs_notation($hgvs_str)->allele_string eq 'C/T', 'HGVSp notat
 ok($vfa->fetch_by_hgvs_notation('ENST00000470094:c.55_111del')->end eq 32954180, 'HGVSc multi-exon deletion');
 ok($vfa->fetch_by_hgvs_notation('NM_000484:c.196_*3del')->end eq 27542743, 'HGVSc multi-exon *deletion');
 
+# test HGVS protein when codon is within two exons
+# forward strand
+ok($vfa->fetch_by_hgvs_notation('ENSP00000422007.1:p.Gly469Glu')->start eq 66325646, 'HGVSp multi-exon (forward)');
+# reverse strand
+ok($vfa->fetch_by_hgvs_notation('ENSP00000293261:p.Arg232Met')->start eq 48846578, 'HGVSp multi-exon (reverse)');
+
+
 print "\n# Test - fetch_by_spdi_notation\n";
 my $spdi_str = 'NC_000013.10:32954017::';
 throws_ok {$vfa->fetch_by_spdi_notation($spdi_str); } qr/Could not parse the SPDI notation $spdi_str/, 'Throw on invalid SPDI notation.';