ENSCORESW-1741: add VGNC source for vertebrate naming

misc-scripts/xref_mapping/XrefParser/HGNCParser.pm

@@ -59,7 +59,9 @@ sub run {
     return 1;
   }
 
-  my $name_to_source_id = $self->get_hgnc_sources();
+  my $source_name = $self->get_source_name_for_source_id($source_id);
+
+  my $name_to_source_id = $self->get_sources($source_name);
 
   # Skip header
   $hugo_io->getline();
@@ -259,15 +261,16 @@ sub run {
 
 
 
-sub get_hgnc_sources {
+sub get_sources {
   my $self = shift;
+  my $source_name = shift;
   my %name_to_source_id;
 
   my @sources = ('entrezgene_manual', 'refseq_manual', 'entrezgene_mapped', 'refseq_mapped', 'ensembl_manual', 'swissprot_manual', 'desc_only');
 
 
   foreach my $key (@sources) {
-  my $source_id = $self->get_source_id_for_source_name('HGNC', $key);
+  my $source_id = $self->get_source_id_for_source_name($source_name, $key);
     if(!(defined $source_id)){
       die 'Could not get source id for HGNC and '. $key ."\n";
     }

misc-scripts/xref_mapping/xref_config.ini

@@ -1102,6 +1102,85 @@ parser          = CGNCParser
 release_uri     =
 data_uri        = http://birdgenenames.org/cgnc/downloads.jsp?file=standard
 
+[source VGNC::MULTI]
+# Used by pan_troglodytes
+name            = VGNC
+download        = Y
+order           = 29
+priority        = 1
+parser          = HGNCParser
+release_uri     =
+data_uri        = file:VGNC_to_Ensembl_mapping.txt.gz
+
+[source VGNC::MULTI#01]
+name            = VGNC
+download        = N
+order           = 30
+priority        = 5
+prio_descr      = refseq_manual
+parser          = HGNCParser
+release_uri     =
+data_uri        =
+
+[source VGNC::MULTI#02]
+name            = VGNC
+download        = N
+order           = 30
+priority        = 5
+prio_descr      = swissprot_manual
+parser          = HGNCParser
+release_uri     =
+data_uri        =
+
+[source VGNC::MULTI#03]
+name            = VGNC
+download        = N
+order           = 30
+priority        = 5
+prio_descr      = ensembl_manual
+parser          = HGNCParser
+release_uri     =
+data_uri        =
+
+[source VGNC::MULTI#04]
+name            = VGNC
+download        = N
+order           = 30
+priority        = 6
+prio_descr      = entrezgene_mapped
+parser          = HGNCParser
+release_uri     =
+data_uri        =
+
+[source VGNC::MULTI#05]
+name            = VGNC
+download        = N
+order           = 30
+priority        = 7
+prio_descr      = refseq_mapped
+parser          = HGNCParser
+release_uri     =
+data_uri        =
+
+[source VGNC::MULTI#06]
+name            = VGNC
+download        = N
+order           = 30
+priority        = 7
+prio_descr      = desc_only
+parser          = HGNCParser
+release_uri     =
+data_uri        =
+
+[source VGNC::MULTI#07]
+name            = VGNC
+download        = N
+order           = 30
+priority        = 6
+prio_descr      = entrezgene_manual
+parser          = HGNCParser
+release_uri     =
+data_uri        =
 
 [source HGNC::homo_sapiens#01]
 # Used by homo_sapiens
@@ -3979,6 +4058,15 @@ prio_descr      =
 parser          = done_in_official_naming
 release_uri     =
 
+[source VGNC_trans_name]
+name            = VGNC_trans_name
+download        = N
+order           = 49
+priority        = 1
+prio_descr      =
+parser          = done_in_official_naming
+release_uri     =
+
 [source HGNC_gene_name]
 name            = HGNC_gene_name
 download        = N
@@ -6570,6 +6658,7 @@ source          = ArrayExpress::MULTI
 [species pan_troglodytes]
 taxonomy_id     = 9598
 aliases         = pantroglodytes, ptroglodytes, chimpanzee, chimp
+source          = VGNC::MULTI
 source          = EntrezGene::MULTI
 source          = GO::MULTI
 source          = Reactome::MULTI