[ENSCORESW-490]. Fixing the GFF serializer which would add an additional ; at...

[ENSCORESW-490]. Fixing the GFF serializer which would add an additional ; at times and now supports a custom source field. Gene and transcript now both emit this

modules/Bio/EnsEMBL/Gene.pm

@@ -1527,6 +1527,7 @@ sub summary_as_hash {
   $summary_ref->{'biotype'} = $self->biotype;
   $summary_ref->{'external_name'} = $self->external_name;
   $summary_ref->{'logic_name'} = $self->analysis->logic_name();
+  $summary_ref->{'source'} = $self->source();
   return $summary_ref;
 }
 

modules/Bio/EnsEMBL/Transcript.pm

@@ -2849,6 +2849,7 @@ sub summary_as_hash {
   $summary_ref->{'logic_name'} = $self->analysis->logic_name();
   my $parent_gene = $self->get_Gene();
   $summary_ref->{'Parent'} = $parent_gene->stable_id;
+  $summary_ref->{'source'} = $self->source() || $parent_gene->source();
   return $summary_ref;
 }
 

modules/Bio/EnsEMBL/Utils/IO/GFFSerializer.pm

@@ -52,6 +52,7 @@ my %strand_conversion = ( '1' => '+', '0' => '?', '-1' => '-');
     Constructor
     Arg [1]    : Ontology Adaptor
     Arg [2]    : Optional File handle
+    Arg [3]    : Default source of the features. Defaults to ensembl
     
     Returntype : Bio::EnsEMBL::Utils::IO::GFFSerializer
 
@@ -62,6 +63,7 @@ sub new {
     my $self = {
         ontology_adaptor => shift,
         filehandle => shift,
+        default_source => shift
     };
     bless $self, $class;
     if ( ! check_ref($self->{'ontology_adaptor'}, "Bio::EnsEMBL::DBSQL::OntologyTermAdaptor" )) {
@@ -74,6 +76,9 @@ sub new {
         open $self->{'filehandle'}, ">&STDOUT";
         $self->{'stdout'} = 1;
     }
+    if(!defined $self->{default_source}) {
+        $self->{default_source} = 'ensembl';
+    }
     return $self;
 }
 
@@ -102,8 +107,9 @@ sub print_feature {
         if (!defined($summary{'seq_region_name'})) {$summary{'seq_region_name'} = "?";}
         $row .= $summary{'seq_region_name'}."\t";
 
-#    Column 2 - source, complicated with Ensembl not being the originator of all data
-        $row .= "EnsEMBL\t";
+#    Column 2 - source, complicated with Ensembl not being the originator of all data but user can specify or it switches to ensembl.
+        $row .= $summary{source} || $self->_default_source();
+        $row .= qq{\t};
 
 #   Column 3 - feature, the ontology term for the kind of feature this row is
         my $so_term = $biotype_mapper->translate_feature_to_SO_term($feature);
@@ -157,6 +163,7 @@ sub print_feature {
         delete $summary{'end'};
         delete $summary{'strand'};
         delete $summary{'score'};
+        delete $summary{'source'};
 #   Slice the hash for specific keys in GFF-friendly order
         my @ordered_keys = qw(ID Name Alias Parent Target Gap Derives_from Note Dbxref Ontology_term Is_circular);
         my @ordered_values = @summary{@ordered_keys};
@@ -168,17 +175,20 @@ sub print_feature {
             delete $summary{$key};
         }
 #   Catch the remaining keys, containing whatever else the Feature provided
-        my @keys = keys %summary;
+        my @keys = sort keys %summary;
         while(my $attribute = shift @keys) {
-            if (ref $summary{$attribute} eq "ARRAY") {
+            my $data_written = 0;
+            if (ref $summary{$attribute} eq "ARRAY" && scalar(@{$summary{$attribute}}) > 0) {
                 $row .= $attribute."=".join (',',map { uri_escape($_,'\t\n\r;=%&,') } grep { defined $_ } @{$summary{$attribute}});
+                $data_written = 1;
             }
             else {
                 if ($summary{$attribute}) { 
                   $row .= $attribute."=".uri_escape($summary{$attribute},'\t\n\r;=%&,'); 
+                  $data_written = 1;
                 }
             }
-            $row .= ';' if scalar(@keys) > 0;
+            $row .= ';' if scalar(@keys) > 0 && $data_written;
         }
 # trim off any trailing commas left by the ordered keys stage above:
         $text_buffer .= $row."\n";
@@ -218,5 +228,10 @@ sub print_metadata {
     print $fh "\n#".$text."\n";
 }
 
+sub _default_source {
+    my ($self) = @_;
+    return $self->{default_source};
+}
+
 
 1;

modules/t/gffSerialiser.t

@@ -42,27 +42,50 @@ my $dba = $db->get_DBAdaptor('core');
 my $id = 'ENSG00000131044';
 
 my $ga = $dba->get_GeneAdaptor();
-my $gene = $ga->fetch_by_stable_id($id);
 
 {
-  my $ota = Test::SO->new();
-  my $fh = IO::String->new();
-  my $ser = Bio::EnsEMBL::Utils::IO::GFFSerializer->new($ota, $fh);
-  $ser->print_main_header([$gene->feature_Slice()]);
-  $ser->print_feature($gene);
-  
+  my $gene = $ga->fetch_by_stable_id($id);
+  delete $gene->{source};
+  $gene->{description} = undef; #empty value means don't emit the key
+  my $expected = <<'OUT';
+##gff-version 3
+##sequence-region   20 30274334 30300924
+OUT
+  #Have to do this outside of the HERETO thanks to tabs
+  $expected .= join("\t", 
+    qw/20  ensembl feature 30274334  30300924  . + ./,
+    'ID=ENSG00000131044;biotype=protein_coding;external_name=C20orf125;logic_name=ensembl' 
+  );
+  $expected .= "\n";
+
+  assert_gff3($gene, $expected, 'Gene with no source serialises to GFF3 as expected. Source is ensembl');
+}
+
+{
+  my $gene = $ga->fetch_by_stable_id($id);
+  $gene->source('wibble');
   my $expected = <<'OUT';
 ##gff-version 3
 ##sequence-region   20 30274334 30300924
 OUT
   #Have to do this outside of the HERETO thanks to tabs
   $expected .= join("\t", 
-    qw/20  EnsEMBL feature 30274334  30300924  . + ./,
-    'ID=ENSG00000131044;logic_name=ensembl;external_name=C20orf125;description=DJ310O13.1.2 (NOVEL PROTEIN SIMILAR DROSOPHILA PROTEIN CG7474%2C ISOFORM 2 ) (FRAGMENT). [Source:SPTREMBL%3BAcc:Q9BR18];biotype=protein_coding' 
+    qw/20  wibble feature 30274334  30300924  . + ./,
+    'ID=ENSG00000131044;biotype=protein_coding;description=DJ310O13.1.2 (NOVEL PROTEIN SIMILAR DROSOPHILA PROTEIN CG7474%2C ISOFORM 2 ) (FRAGMENT). [Source:SPTREMBL%3BAcc:Q9BR18];external_name=C20orf125;logic_name=ensembl' 
   );
   $expected .= "\n";
 
-  is(${$fh->string_ref()}, $expected, 'Gene serialises to GFF3 as expected');
+  assert_gff3($gene, $expected, 'Gene with custom source serialises to GFF3 as expected. Source is wibble');
+}
+
+sub assert_gff3 {
+  my ($feature, $expected, $msg) = @_;
+  my $ota = Test::SO->new();
+  my $fh = IO::String->new();
+  my $ser = Bio::EnsEMBL::Utils::IO::GFFSerializer->new($ota, $fh);
+  $ser->print_main_header([$feature->feature_Slice()]);
+  $ser->print_feature($feature);
+  is(${$fh->string_ref()}, $expected, $msg);
 }
 
 done_testing();
\ No newline at end of file