fix for multiple custom annotations at same locus

Ensembl · Feb 20, 2017 · 11c3214 · 11c3214
1 parent e79ac34
commit 11c3214
Show file tree

Hide file tree

Showing 5 changed files with 90 additions and 6 deletions.
diff --git a/modules/Bio/EnsEMBL/VEP/Constants.pm b/modules/Bio/EnsEMBL/VEP/Constants.pm
@@ -20,7 +20,7 @@ use warnings;
 use base qw(Exporter);
 
 our $VEP_VERSION     = 87;
-our $VEP_SUB_VERSION = 24;
+our $VEP_SUB_VERSION = 25;
 
 our @EXPORT_OK = qw(
   @FLAG_FIELDS

diff --git a/modules/Bio/EnsEMBL/VEP/OutputFactory.pm b/modules/Bio/EnsEMBL/VEP/OutputFactory.pm
@@ -806,10 +806,10 @@ sub _add_custom_annotations_to_hash {
   my ($self, $hash, $custom_name, $annots) = @_;
 
   foreach my $annot(@$annots) {
-    $hash->{$custom_name} = $annot->{name};
+    push @{$hash->{$custom_name}}, $annot->{name};
 
     foreach my $field(keys %{$annot->{fields} || {}}) {
-      $hash->{$custom_name.'_'.$field} = $annot->{fields}->{$field};
+      push @{$hash->{$custom_name.'_'.$field}}, $annot->{fields}->{$field};
     }
   }
 

diff --git a/t/OutputFactory.t b/t/OutputFactory.t
@@ -240,7 +240,25 @@ is_deeply(
 );
 delete($of->{individual});
 
-$of->get_all_output_hashes_by_VariationFeature($ib->buffer->[0]);
+
+$ib->buffer->[0]->{_custom_annotations} = {
+  custom1 => [ {name => 'foo'} ],
+  custom2 => [ {name => 'bar'}, {name => 'car'} ],
+  custom3 => [ {allele => 'A', name => 'shoe'} ], # this one will be ignored as it has allele key
+};
+is_deeply(
+  $of->VariationFeature_to_output_hash($ib->buffer->[0]),
+  {
+    'Uploaded_variation' => 'indtest',
+    'Location' => '21:25587759',
+    'IND' => 'dave',
+    'ZYG' => 'HET',
+    'custom1' => ['foo'],
+    'custom2' => ['bar', 'car'],
+  },
+  'VariationFeature_to_output_hash - custom annotations'
+);
+
 
 
 
@@ -589,6 +607,27 @@ $of->{flag_pick} = 0;
 
 
 
+$ib->buffer->[0]->{_custom_annotations} = {
+  custom1 => [ {allele => 'T', name => 'foo'} ],
+  custom2 => [ {name => 'bar'}, {name => 'car'} ], # this one will be ignored as it has no allele key
+  custom3 => [ {allele => 'T', name => 'shoe'}, {allele => 'T', name => 'moo'} ],
+};
+is_deeply(
+  $of->VariationFeatureOverlapAllele_to_output_hash($vfoa, {}, $ib->buffer->[0]),
+  {
+    'IMPACT' => 'MODIFIER',
+    'Consequence' => [
+      '3_prime_UTR_variant'
+    ],
+    'Allele' => 'T',
+    'PICK' => 1,
+    'custom1' => ['foo'],
+    'custom3' => ['shoe', 'moo'],
+  },
+  'VariationFeatureOverlapAllele_to_output_hash - custom annotations'
+);
+
+
 
 ## frequency tests
 ##################

diff --git a/t/OutputFactory_JSON.t b/t/OutputFactory_JSON.t
@@ -462,7 +462,7 @@ SKIP: {
     no warnings 'once';
 
     ## REMEMBER TO UPDATE THIS SKIP NUMBER IF YOU ADD MORE TESTS!!!!
-    skip 'Bio::DB::HTS::Tabix module not available', 1 unless $Bio::EnsEMBL::VEP::AnnotationSource::File::CAN_USE_TABIX_PM;
+    skip 'Bio::DB::HTS::Tabix module not available', 2 unless $Bio::EnsEMBL::VEP::AnnotationSource::File::CAN_USE_TABIX_PM;
 
     $ib = get_annotated_buffer({
       input_file => $test_cfg->{test_vcf},
@@ -488,6 +488,34 @@ SKIP: {
       },
       'custom_annotations'
     );
+
+    $ib = get_annotated_buffer({
+      input_data => "21\t25585733\t.\tCATG\tTACG",
+      everything => 1,
+      dir => $test_cfg->{cache_root_dir},
+      custom => [$test_cfg->{custom_vcf}.',test,vcf,overlap'],
+    });
+    $of = Bio::EnsEMBL::VEP::OutputFactory::JSON->new({config => $ib->config});
+    @lines = @{$of->get_all_lines_by_InputBuffer($ib)};
+
+    is_deeply(
+      $json->decode($lines[0])->{custom_annotations},
+      {
+        "test" => [
+          {
+            "name" => "test1",
+          },
+          {
+            "name" => "del1",
+          },
+          {
+            "name" => "del2",
+          }
+
+        ]
+      },
+      'custom_annotations overlap'
+    );
   }
 }
 

diff --git a/t/OutputFactory_VCF.t b/t/OutputFactory_VCF.t
@@ -356,7 +356,7 @@ SKIP: {
   no warnings 'once';
 
   ## REMEMBER TO UPDATE THIS SKIP NUMBER IF YOU ADD MORE TESTS!!!!
-  skip 'Bio::DB::HTS::Tabix module not available', 1 unless $Bio::EnsEMBL::VEP::AnnotationSource::File::CAN_USE_TABIX_PM;
+  skip 'Bio::DB::HTS::Tabix module not available', 2 unless $Bio::EnsEMBL::VEP::AnnotationSource::File::CAN_USE_TABIX_PM;
 
   $runner = get_runner({
     input_file => $test_cfg->{test_vcf},
@@ -373,6 +373,23 @@ SKIP: {
     "T|upstream_gene_variant|MODIFIER||ENSG00000260583|Transcript|ENST00000567517||||||||||||2407|-1|||test1|BAR\tGT\t0|0",
     'get_all_lines_by_InputBuffer - custom'
   );
+
+
+  $runner = get_runner({
+    input_data => "21\t25585733\t.\tCATG\tTACG",
+    custom => [$test_cfg->{custom_vcf}.',test,vcf,overlap'],
+    output_format => 'vcf',
+  });
+  $of = $runner->get_OutputFactory;
+
+  is(
+    $of->get_all_lines_by_InputBuffer($runner->get_InputBuffer)->[0],
+    "21\t25585733\t.\tCATG\tTACG\t.\t.\t".
+    "CSQ=TACG|3_prime_UTR_variant|MODIFIER||ENSG00000154719|Transcript|ENST00000307301||||||1119-1122|||||||-1|||test1&del1&del2,".
+    "TACG|missense_variant|MODERATE||ENSG00000154719|Transcript|ENST00000352957||||||1030-1033|988-991|330-331|HA/RT|CATGca/CGTAca|||-1|||test1&del1&del2,".
+    "TACG|upstream_gene_variant|MODIFIER||ENSG00000260583|Transcript|ENST00000567517||||||||||||2407|-1|||test1&del1&del2",
+    'get_all_lines_by_InputBuffer - custom overlap'
+  );
 }
 
 # test converting to VCF from different input