From 74c61732ea51220d7de84874ee46149421cb8cff Mon Sep 17 00:00:00 2001
From: Magali Ruffier <mr6@ebi.ac.uk>
Date: Thu, 8 Feb 2018 09:14:09 +0000
Subject: [PATCH] custom download for HGNC
---
.../xref_mapping/XrefParser/HGNCParser.pm | 23 ++++++++-----------
misc-scripts/xref_mapping/xref_config.ini | 2 +-
2 files changed, 11 insertions(+), 14 deletions(-)
diff --git a/misc-scripts/xref_mapping/XrefParser/HGNCParser.pm b/misc-scripts/xref_mapping/XrefParser/HGNCParser.pm
index f77cb949d4..39f1989c66 100644
--- a/misc-scripts/xref_mapping/XrefParser/HGNCParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/HGNCParser.pm
@@ -142,14 +142,11 @@ sub run_script {
my $acc = $array[0];
my $symbol = $array[1];
my $name = $array[2];
- my $status = $array[5];
- my $previous_symbols = $array[8];
- my $synonyms = $array[10];
-
- if ($status ne 'Approved') { next; }
+ my $previous_symbols = $array[3];
+ my $synonyms = $array[4];
my $type = 'lrg';
- my $id = $array[29];
+ my $id = $array[9];
my $source_id = $name_to_source_id->{$type};
if($id and $id =~ m/http:\/\/www.lrg-sequence.org\/LRG\/(LRG_\d+)/x){
my $lrg_stable_id = $1;
@@ -178,9 +175,9 @@ sub run_script {
$type = 'ccds';
$source_id = $name_to_source_id->{$type};
- my $ccds = $array[24];
+ my $ccds = $array[8];
$ccds =~ s/"//g if defined $ccds;
- my @ccds_list = split(/\|/,$ccds) if defined $ccds;
+ my @ccds_list = split(/,\s/,$ccds) if defined $ccds;
foreach my $ccds (@ccds_list) {
$id = $ccds_to_ens{$ccds};
@@ -208,7 +205,7 @@ sub run_script {
# Direct Ensembl mappings
#
$type = 'ensembl_manual';
- $id = $array[19];
+ $id = $array[6];
$source_id = $name_to_source_id->{$type};
if ($id){ # Ensembl direct xref
$seen = 1;
@@ -232,7 +229,7 @@ sub run_script {
}
$type = 'refseq_manual';
- $id = $array[23];
+ $id = $array[7];
$source_id = $name_to_source_id->{$type};
if ($id) {
if(defined $refseq{$id} ){
@@ -260,7 +257,7 @@ sub run_script {
# EntrezGene
#
$type = 'entrezgene_manual';
- $id = $array[18];
+ $id = $array[5];
$source_id = $name_to_source_id->{$type};
if(defined $id ){
if(defined $entrezgene{$id} ){
@@ -353,7 +350,7 @@ sub add_synonyms_for_hgnc{
my $dbi = $ref_arg->{dbi};
if (defined $dead_name ) { # dead name, add to synonym
- my @array2 = split '\|', $dead_name ;
+ my @array2 = split ',\s', $dead_name ;
foreach my $arr (@array2){
$arr =~ s/"//g;
$self->add_to_syn($name, $source_id, $arr, $species_id, $dbi);
@@ -361,7 +358,7 @@ sub add_synonyms_for_hgnc{
}
if (defined $alias ) { # alias, add to synonym
- my @array2 = split '\|', $alias;
+ my @array2 = split ',\s', $alias;
foreach my $arr (@array2){
$arr =~ s/"//g;
$self->add_to_syn($name, $source_id, $arr, $species_id, $dbi);
diff --git a/misc-scripts/xref_mapping/xref_config.ini b/misc-scripts/xref_mapping/xref_config.ini
index 15338485df..a452773515 100644
--- a/misc-scripts/xref_mapping/xref_config.ini
+++ b/misc-scripts/xref_mapping/xref_config.ini
@@ -1196,7 +1196,7 @@ prio_descr = entrezgene_manual
parser = HGNCParser
dependent_on = EntrezGene,Uniprot/SWISSPROT,RefSeq_dna,RefSeq_peptide
release_uri =
-data_uri = script:wget=>ftp://ftp.ebi.ac.uk/pub/databases/genenames/new/tsv/hgnc_complete_set.txt,host=>mysql-ens-core-prod-1,dbname=>homo_sapiens_ccds_91_38,port=>4524,
+data_uri = script:wget=>https://www.genenames.org/cgi-bin/download?col=gd_hgnc_id&col=gd_app_sym&col=gd_app_name&col=gd_prev_sym&col=gd_aliases&col=gd_pub_eg_id&col=gd_pub_ensembl_id&col=gd_pub_refseq_ids&col=gd_ccds_ids&col=gd_lsdb_links&status=Approved&status_opt=2&where=&order_by=gd_app_sym_sort&format=text&limit=&hgnc_dbtag=on&submit=submit,host=>mysql-ens-core-prod-1,dbname=>homo_sapiens_ccds_91_38,port=>4524,
[source HGNC::homo_sapiens#03]
--
GitLab