From 980bdb4d212211119afee9261de5389299f4af74 Mon Sep 17 00:00:00 2001
From: Magali Ruffier <mr6@ebi.ac.uk>
Date: Fri, 3 Nov 2017 16:33:37 +0000
Subject: [PATCH] capture mRNA for peptide information

---
 misc-scripts/xref_mapping/XrefParser/RefSeqGPFFParser.pm | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/misc-scripts/xref_mapping/XrefParser/RefSeqGPFFParser.pm b/misc-scripts/xref_mapping/XrefParser/RefSeqGPFFParser.pm
index b550cfbc44..37da140c57 100644
--- a/misc-scripts/xref_mapping/XrefParser/RefSeqGPFFParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/RefSeqGPFFParser.pm
@@ -202,6 +202,7 @@ sub create_xrefs {
     {
       my ($acc) = $entry =~ /ACCESSION\s+(\S+)/;
       my ($ver) = $entry =~ /VERSION\s+(\S+)/;
+      my ($refseq_pair) = $entry =~ /DBSOURCE\s+REFSEQ: accession (\S+)/;
 
       # get the right source ID based on $type and whether this is predicted (X*) or not
       my $source_id;
@@ -272,6 +273,9 @@ sub create_xrefs {
       foreach my $cb (@coded_by){
 	$xref->{PAIR} = $cb;
       }
+      if (!defined $xref->{PAIR}) {
+        $xref->{PAIR} = $refseq_pair;
+      }
 
       foreach my $pi (@protein_id){
 	$xref->{PROTEIN} = $pi;
-- 
GitLab