Merge pull request #49 from pcantalupo/topic/bug-3375

Fixed bug 3375 and added some tests
bioperl · Nov 7, 2012 · 8039fe1 · 8039fe1
2 parents fac6f33 + 65bb304
commit 8039fe1
Show file tree

Hide file tree

Showing 3 changed files with 517 additions and 11 deletions.
diff --git a/Bio/SeqIO/genbank.pm b/Bio/SeqIO/genbank.pm
@@ -587,20 +587,30 @@ sub next_seq {
                       -version => $version,
                       -database => $db || 'GenBank',
                       -tagname => 'dblink'));
-                } elsif ( $dbsource =~ /(\S+)([\.:])\s*(\d+)/ ) {
-                    my ($id, $db, $version);
+                } elsif ( $dbsource =~ /(\S+)([\.:])\s*(\S+)/ ) {
+                    my ($db, $version);
+                    my @ids = ();
                     if ($2 eq ':') {
-                        ($db, $id) = ($1, $3);
+                        $db = $1;
+                        # Genbank 192 release notes say this: "The second field can consist of
+                        #     multiple comma-separated identifiers, if a sequence record has
+                        #     multiple DBLINK cross-references of a given type."
+                        #     For example: DBLINK      Project:100,200,300"
+                        @ids = split (/,/, $3);
                     } else {
-                        ($db, $id, $version) = ('GenBank', $1, $3);
+                        ($db, $version) = ('GenBank', $3);
+                        $ids[0] = $1;
                     }
-                    $annotation->add_Annotation('dblink',
-                        Bio::Annotation::DBLink->new(
-                            -primary_id => $id,
-                            -version => $version,
-                            -database => $db,
-                            -tagname => 'dblink')
+
+                    foreach my $id (@ids) {
+                        $annotation->add_Annotation('dblink',
+                            Bio::Annotation::DBLink->new(
+                               -primary_id => $id,
+                               -version => $version,
+                               -database => $db,
+                               -tagname => 'dblink')
                         );
+                    }
                 } else {
                     $self->warn("Unrecognized DBSOURCE data: $dbsource\n");
                 }

diff --git a/t/SeqIO/genbank.t b/t/SeqIO/genbank.t
@@ -7,7 +7,7 @@ BEGIN {
     use lib '.';
 	use Bio::Root::Test;
 
-	test_begin(-tests => 274 );
+	test_begin(-tests => 281 );
 
     use_ok('Bio::SeqIO::genbank');
 }
@@ -601,3 +601,20 @@ $as = $ast->next_seq;
 ($cds) = grep { $_->primary_tag eq 'CDS' } $as->get_SeqFeatures();
 @notes = $cds->get_tag_values('note');
 is(scalar @notes, 2);
+
+
+#bug 3375
+my $in = Bio::SeqIO->new(-format => 'genbank',
+                        -file => test_input_file('NC_002058_multDBLINK_bug3375.gb'));
+my $seq = $in->next_seq();     # should not throw a warning now
+my @dblinks = $seq->annotation->get_Annotations('dblink');    # contains 5 dblink references
+# testing DBLINK      BioProject: PRJNA15288
+is($dblinks[0]->database, 'BioProject', 'bug3375 database is BioProject');
+is($dblinks[0]->primary_id, 'PRJNA15288', 'bug3375 primary_id is PRJNA15288');
+# testing DBLINK      Project:100,200,300
+is($dblinks[3]->database, 'Project');
+is($dblinks[3]->primary_id, '300');
+# testing DBLINK      NC_002058.3
+is($dblinks[4]->database, 'GenBank');
+is($dblinks[4]->primary_id, 'NC_002058');
+is($dblinks[4]->version, '3');