From 39f0a770ac136e6c9f6e935739844f5cb604021d Mon Sep 17 00:00:00 2001
From: Magali Ruffier <mr6@ebi.ac.uk>
Date: Thu, 2 Oct 2014 16:01:30 +0100
Subject: [PATCH] first pass CGNCParser for chicken names recovered from cvs
commits in 2012, not tested
---
.../xref_mapping/XrefMapper/gallus_gallus.pm | 66 +++++++++++++++++++
.../xref_mapping/XrefParser/CGNCParser.pm | 65 ++++++++++++++++++
misc-scripts/xref_mapping/xref_config.ini | 13 ++++
3 files changed, 144 insertions(+)
create mode 100644 misc-scripts/xref_mapping/XrefMapper/gallus_gallus.pm
create mode 100644 misc-scripts/xref_mapping/XrefParser/CGNCParser.pm
diff --git a/misc-scripts/xref_mapping/XrefMapper/gallus_gallus.pm b/misc-scripts/xref_mapping/XrefMapper/gallus_gallus.pm
new file mode 100644
index 0000000000..b0ca583cd3
--- /dev/null
+++ b/misc-scripts/xref_mapping/XrefMapper/gallus_gallus.pm
@@ -0,0 +1,66 @@
+package XrefMapper::gallus_gallus;
+
+use XrefMapper::BasicMapper;
+
+use vars '@ISA';
+
+@ISA = qw{ XrefMapper::BasicMapper };
+use strict;
+
+
+sub gene_display_xref_sources {
+ my $self = shift;
+
+ my @list = qw(CGNC
+ RFAM
+ miRBase
+ Uniprot_gn
+ EntrezGene);
+
+ my %ignore;
+
+ #don't use EntrezGene labels dependent on predicted RefSeqs
+
+$ignore{'EntrezGene'} =<<IEG;
+SELECT DISTINCT ox.object_xref_id
+ FROM object_xref ox, dependent_xref dx,
+ xref xmas, xref xdep,
+ source smas, source sdep
+ WHERE ox.xref_id = dx.dependent_xref_id AND
+ dx.dependent_xref_id = xdep.xref_id AND
+ dx.master_xref_id = xmas.xref_id AND
+ xmas.source_id = smas.source_id AND
+ xdep.source_id = sdep.source_id AND
+ smas.name like "Refseq%predicted" AND
+ sdep.name like "EntrezGene" AND
+ ox.ox_status = "DUMP_OUT"
+IEG
+
+ #don't use labels starting with LOC
+
+$ignore{'LOC_prefix'} =<<LOCP;
+SELECT object_xref_id
+ FROM object_xref JOIN xref USING(xref_id) JOIN source USING(source_id)
+ WHERE ox_status = 'DUMP_OUT' AND label REGEXP '^LOC[[:digit:]]+'
+LOCP
+
+ return [\@list,\%ignore];
+
+}
+
+
+sub gene_description_sources {
+
+ return ("miRBase",
+ "RFAM",
+ "CGNC",
+ "Uniprot/SWISSPROT",
+ "Uniprot/Varsplic",
+ "RefSeq_peptide",
+ "RefSeq_mRNA",
+ "Uniprot/SPTREMBL" );
+
+}
+
+
+1;
diff --git a/misc-scripts/xref_mapping/XrefParser/CGNCParser.pm b/misc-scripts/xref_mapping/XrefParser/CGNCParser.pm
new file mode 100644
index 0000000000..818b28806a
--- /dev/null
+++ b/misc-scripts/xref_mapping/XrefParser/CGNCParser.pm
@@ -0,0 +1,65 @@
+package XrefParser::CGNCParser;
+
+use strict;
+use warnings;
+use Carp;
+use DBI;
+
+use base qw(XrefParser::BaseParser);
+
+sub run {
+
+ my ($self, $ref_arg) = @_;
+ my $source_id = $ref_arg->{source_id};
+ my $species_id = $ref_arg->{species_id};
+ my $files = $ref_arg->{files};
+ my $verbose = $ref_arg->{verbose};
+
+ if((!defined $source_id) or (!defined $species_id) or (!defined $files) ){
+ croak "Need to pass source_id, species_id and files as pairs";
+ }
+ $verbose |=0;
+
+ my $file = @{$files}[0];
+
+ my $file_io = $self->get_filehandle($file);
+ if ( !defined $file_io ) {
+ print STDERR "ERROR: Could not open $file\n";
+ return 1; # 1 is an error
+ }
+
+ $source_id = $self->get_source_id_for_source_name("CGNC");
+
+
+ my $count = 0;
+ while ( $_ = $file_io->getline() ) {
+#48941 ENSGALG00000002652 FZD10
+
+ chomp;
+ my @array = split /\t/x, $_;
+
+ my $ensid = $array[1];
+ my $acc = $array[2];
+ my $desc = $array[3];
+
+ if($ensid =~ /ENSGAL/){
+ my $xref_id = $self->add_xref({ acc => $acc,
+ version => 0,
+ label => $acc,
+ desc => $desc,
+ source_id => $source_id,
+ species_id => $species_id,
+ info_type => "DIRECT"} );
+
+ $self->add_direct_xref( $xref_id, $ensid, "Gene", '');
+ $count++;
+ } else{
+ print STDERR "No match for $acc\n";
+ }
+ }
+ print "$count direct CGNC xrefs added\n";
+ return 0;
+
+}
+
+1;
diff --git a/misc-scripts/xref_mapping/xref_config.ini b/misc-scripts/xref_mapping/xref_config.ini
index 623927544d..d1ec7b52b6 100644
--- a/misc-scripts/xref_mapping/xref_config.ini
+++ b/misc-scripts/xref_mapping/xref_config.ini
@@ -1129,6 +1129,18 @@ parser = HGNCParser
release_uri =
data_uri =
+[source CGNC::gallus_gallus]
+# Used by chicken
+name = CGNC
+download = Y
+order = 30
+priority = 1
+prio_descr = cgnc
+parser = CGNCParser
+release_uri =
+data_uri = script:wget=>http://birdgenenames.org/cgnc/downloads.jsp,
+
+
[source HGNC::homo_sapiens#01]
# Used by homo_sapiens
name = HGNC
@@ -6050,6 +6062,7 @@ source = UniParc::MULTI
source = RFAM::MULTI
source = miRBase::MULTI
source = ArrayExpress::MULTI
+source = CGNC::gallus_gallus
[species gasterosteus_aculeatus]
taxonomy_id = 69293
--
GitLab