[ENSCORESW-438]. Fixed API when someone can specify a _rna_edit where the...

[ENSCORESW-438]. Fixed API when someone can specify a _rna_edit where the substring length was not equal to the substituting string length

modules/Bio/EnsEMBL/DBSQL/SequenceAdaptor.pm

@@ -88,8 +88,7 @@ sub new {
 # See if this has any seq_region_attrib of type "_rna_edit_cache" if so store these
 # in a  hash.
 #
-
-  my $sth = $self->dbc->prepare('select sra.seq_region_id, sra.value from seq_region_attrib sra, attrib_type at where sra.attrib_type_id = at.attrib_type_id and code like "_rna_edit"');
+  my $sth = $self->dbc->prepare('select sra.seq_region_id, sra.value from seq_region_attrib sra, attrib_type at where sra.attrib_type_id = at.attrib_type_id and code = "_rna_edit"');
   
   $sth->execute();
   my ($seq_region_id, $value);
@@ -98,6 +97,12 @@ sub new {
   my $count = 0;
   while($sth->fetch()){
     $count++;
+    my ($start, $end, $substring) = split (/\s+/, $value);
+    my $edit_length = ($end - $start) + 1;
+    my $substring_length = length($substring);
+    if($edit_length != $substring_length) {
+      throw "seq_region_id $seq_region_id has an attrib of type '_rna_edit' (value '$value'). Edit length ${edit_length} is not the same as the replacement's length ${substring_length}. Please fix. We only support substitutions via this mechanism";
+    }
     push @{$edits{$seq_region_id}}, $value;
   }
   $sth->finish;

modules/t/sequenceAdaptor.t

@@ -2,11 +2,13 @@ use strict;
 use warnings;
 
 use Test::More;
+use Test::Exception;
 
 use Bio::EnsEMBL::Test::TestUtils;
 
 use Bio::EnsEMBL::Test::MultiTestDB;
 use Bio::EnsEMBL::Slice;
+use Bio::EnsEMBL::Attribute;
 
 our $verbose= 0;
 
@@ -26,22 +28,37 @@ my $STRAND        = 1;
 my $slice_adaptor = $db->get_SliceAdaptor;
 my $seq_adaptor = $db->get_SequenceAdaptor();
 
+{
+  my $slice = $slice_adaptor->fetch_by_region('chromosome', $CHR, $START, $END);
+  compare_compliments($slice, $seq_adaptor);
 
-my $slice = $slice_adaptor->fetch_by_region('chromosome', $CHR, $START, $END);
-compare_compliments($slice, $seq_adaptor);
+  #Bigger than 1Mb
+  $slice = $slice_adaptor->fetch_by_region('chromosome', $CHR, $START, $START+2_000_000);
+  compare_compliments($slice, $seq_adaptor);
 
-#Bigger than 1Mb
-$slice = $slice_adaptor->fetch_by_region('chromosome', $CHR, $START, $START+2_000_000);
-compare_compliments($slice, $seq_adaptor);
+  $slice = $slice_adaptor->fetch_by_region('clone','AL031658.11');
+  compare_compliments($slice, $seq_adaptor);
 
-$slice = $slice_adaptor->fetch_by_region('clone','AL031658.11');
-compare_compliments($slice, $seq_adaptor);
+  $slice = $slice_adaptor->fetch_by_region('supercontig','NT_028392');
+  compare_compliments($slice, $seq_adaptor);
 
-$slice = $slice_adaptor->fetch_by_region('supercontig','NT_028392');
-compare_compliments($slice, $seq_adaptor);
+  $slice = $slice_adaptor->fetch_by_region('contig', 'AL031658.11.1.162976');
+  compare_compliments($slice, $seq_adaptor);
+}
 
-$slice = $slice_adaptor->fetch_by_region('contig', 'AL031658.11.1.162976');
-compare_compliments($slice, $seq_adaptor);
+$multi_db->save('core','seq_region_attrib');
+{
+  #Adding an insertion _rna_edit into the seq_region_attrib table. This should
+  #not occur and we should fail to build the adaptor whilst it is still there.
+  #We will attach it to the clone
+  my $attribute_adaptor = $db->get_AttributeAdaptor();
+  my $bad_attribute = Bio::EnsEMBL::Attribute->new(-CODE => '_rna_edit', -VALUE => '10 10 AT'); # 1bp edit but string is 2bp
+  my $slice = $slice_adaptor->fetch_by_region('clone','AL031658.11');
+  $attribute_adaptor->store_on_Slice($slice, [$bad_attribute]);
+
+  throws_ok { $seq_adaptor->new($db); } qr/Edit length .+ substitutions/, 'Exception will be thrown when attempting to build a SequenceAdaptor with a mis-match edit length _rna_edit (only support subs)';
+}
+$multi_db->restore('core','seq_region_attrib');
 
 
 sub compare_compliments {
@@ -50,13 +67,13 @@ sub compare_compliments {
 
   my $seq = ${$seq_adaptor->fetch_by_Slice_start_end_strand($slice,1,undef,1)};
 
-  debug('FORWARD STRAND SLICE SEQ for ' . $slice->name());
-  debug($slice->length);
+  note('FORWARD STRAND SLICE SEQ for ' . $slice->name());
+  note($slice->length);
 
   my $invert_seq = 
     ${$seq_adaptor->fetch_by_Slice_start_end_strand($slice->invert,1,undef,1)};
 
-  debug('REVERSE STRAND SLICE SEQ for ' . $slice->name());
+  note('REVERSE STRAND SLICE SEQ for ' . $slice->name());
   
 
   is(length($seq), $slice->length, 'sequence is correct length');
@@ -64,7 +81,12 @@ sub compare_compliments {
   $seq = reverse $seq;  #reverse complement seq
   $seq =~ tr/ACTG/TGAC/;
 
-  ok($seq eq $invert_seq, 'revcom same as seq on inverted slice');
+  #Only use ok here; is would just be crazy
+  my $ok = ok($seq eq $invert_seq, 'revcom same as seq on inverted slice');
+  if(! $ok ) {
+    diag 'SEQ:    '.$seq;
+    diag 'INVERT: '.$invert_seq;
+  }
 }
 
 done_testing();