fix range-finding error for insertions when not using Set::IntervalTree

Ensembl · Jun 22, 2017 · 95350a9 · 95350a9
1 parent 99cbe3f
commit 95350a9
Show file tree

Hide file tree

Showing 3 changed files with 86 additions and 2 deletions.
diff --git a/modules/Bio/EnsEMBL/VEP/Constants.pm b/modules/Bio/EnsEMBL/VEP/Constants.pm
@@ -53,7 +53,7 @@ use warnings;
 use base qw(Exporter);
 
 our $VEP_VERSION     = 89;
-our $VEP_SUB_VERSION = 3;
+our $VEP_SUB_VERSION = 4;
 
 our @EXPORT_OK = qw(
   @FLAG_FIELDS

diff --git a/modules/Bio/EnsEMBL/VEP/InputBuffer.pm b/modules/Bio/EnsEMBL/VEP/InputBuffer.pm
@@ -247,7 +247,10 @@ sub get_overlapping_vfs {
     my $hash_tree = $self->hash_tree;
 
     return [
-      grep {overlap($_->{start}, $_->{end}, $start, $end)}
+      grep {  # checking both overlaps is quicker than sorting here
+        overlap($_->{start}, $_->{end}, $start, $end) ||
+        overlap($_->{end}, $_->{start}, $start, $end)
+      }
       values %{{
         map {$_->{_hash_tree_id} => $_}                  # use _hash_tree_id to uniquify
         map {@{$hash_tree->{$_} || []}}                  # tree might be empty

diff --git a/t/InputBuffer.t b/t/InputBuffer.t
@@ -184,6 +184,45 @@ is_deeply(
   'get_overlapping_vfs - big SV 2'
 );
 
+# insertion between 11 and 12
+$ib->buffer->[0]->{start}++;
+
+# reset trees
+delete $ib->{temp}->{interval_tree};
+delete $ib->{temp}->{hash_tree};
+
+is_deeply(
+  $ib->get_overlapping_vfs(25592910, 25592911),
+  [$ib->buffer->[0]],
+  'get_overlapping_vfs - ins 1'
+);
+
+is_deeply(
+  $ib->get_overlapping_vfs(25592912, 25592913),
+  [$ib->buffer->[0]],
+  'get_overlapping_vfs - ins 2'
+);
+
+is_deeply(
+  $ib->get_overlapping_vfs(25592911, 25592912),
+  [$ib->buffer->[0]],
+  'get_overlapping_vfs - ins 3'
+);
+
+is_deeply(
+  $ib->get_overlapping_vfs(25592909, 25592910),
+  [],
+  'get_overlapping_vfs - ins 4'
+);
+
+is_deeply(
+  $ib->get_overlapping_vfs(25592913, 25592914),
+  [],
+  'get_overlapping_vfs - ins 5'
+);
+
+$ib->buffer->[0]->{start}--;
+
 SKIP: {
 
   ## REMEMBER TO UPDATE THIS SKIP NUMBER IF YOU ADD MORE TESTS!!!!
@@ -194,6 +233,9 @@ SKIP: {
   $Bio::EnsEMBL::VEP::InputBuffer::CAN_USE_INTERVAL_TREE = 0;
 
   delete $ib->{temp}->{interval_tree};
+  delete $ib->{temp}->{hash_tree};
+
+  $DB::single = 1;
 
   is_deeply(
     $ib->get_overlapping_vfs(25592911, 25592911),
@@ -255,6 +297,45 @@ SKIP: {
     'get_overlapping_vfs no tree - big SV 2'
   );
 
+  # insertion between 11 and 12
+  $ib->buffer->[0]->{start}++;
+
+  # reset trees
+  delete $ib->{temp}->{interval_tree};
+  delete $ib->{temp}->{hash_tree};
+
+  is_deeply(
+    $ib->get_overlapping_vfs(25592910, 25592911),
+    [$ib->buffer->[0]],
+    'get_overlapping_vfs no tree - ins 1'
+  );
+
+  is_deeply(
+    $ib->get_overlapping_vfs(25592912, 25592913),
+    [$ib->buffer->[0]],
+    'get_overlapping_vfs no tree - ins 2'
+  );
+
+  is_deeply(
+    $ib->get_overlapping_vfs(25592911, 25592912),
+    [$ib->buffer->[0]],
+    'get_overlapping_vfs no tree - ins 3'
+  );
+
+  is_deeply(
+    $ib->get_overlapping_vfs(25592909, 25592910),
+    [],
+    'get_overlapping_vfs no tree - ins 4'
+  );
+
+  is_deeply(
+    $ib->get_overlapping_vfs(25592913, 25592914),
+    [],
+    'get_overlapping_vfs no tree - ins 5'
+  );
+
+  $ib->buffer->[0]->{start}--;
+
   $Bio::EnsEMBL::VEP::InputBuffer::CAN_USE_INTERVAL_TREE = $orig;
 }