From fce7e886bf0cbdb455710da9719eea9141e4ccbb Mon Sep 17 00:00:00 2001
From: Monika Komorowska <mk8@sanger.ac.uk>
Date: Mon, 11 Jun 2012 15:07:29 +0000
Subject: [PATCH] another attempt to accommodate multiple ensembl stable ids
linking to the same vega stable id
---
.../XrefParser/curated_transcriptParser.pm | 20 ++++++++++---------
1 file changed, 11 insertions(+), 9 deletions(-)
diff --git a/misc-scripts/xref_mapping/XrefParser/curated_transcriptParser.pm b/misc-scripts/xref_mapping/XrefParser/curated_transcriptParser.pm
index c9fbe1d04f..0165622ded 100644
--- a/misc-scripts/xref_mapping/XrefParser/curated_transcriptParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/curated_transcriptParser.pm
@@ -156,25 +156,27 @@ sub run_script {
print "source id is $source_id, curated_source_id is $curated_source_id\n";
- my $sql = 'select t.stable_id, x.display_label, t.status from analysis a, xref x, object_xref ox , external_db e, transcript t where t.analysis_id = a.analysis_id and a.logic_name like "%havana%" and e.external_db_id = x.external_db_id and x.xref_id = ox.xref_id and t.transcript_id = ox.ensembl_id and e.db_name like ?';
+ my $sql = 'select distinct t.stable_id, x.display_label, t.status from analysis a, xref x, object_xref ox , external_db e, transcript t where t.analysis_id = a.analysis_id and a.logic_name like "%havana%" and e.external_db_id = x.external_db_id and x.xref_id = ox.xref_id and t.transcript_id = ox.ensembl_id and e.db_name like ?';
my $sql_vega = 'select t.stable_id, x.display_label, t.status from xref x, object_xref ox , external_db e, transcript t where e.external_db_id = x.external_db_id and x.xref_id = ox.xref_id and t.transcript_id = ox.ensembl_id and t.stable_id <> x.display_label and e.db_name like ?';
my %ott_to_vega_name;
- my %enst_to_ott;
+ my %ott_to_enst;
my $sth = $core_dbc->prepare($sql) || die "Could not prepare for core $sql\n";
+ my $count_ott_to_enst;
foreach my $external_db (qw(Vega_transcript shares_CDS_with_OTTT shares_CDS_and_UTR_with_OTTT OTTT)){
$sth->execute($external_db) or croak( $core_dbc->errstr());
while ( my @row = $sth->fetchrow_array() ) {
- $enst_to_ott{$row[0]} = $row[1];
+ push (@{$ott_to_enst{$row[1]}},$row[0]);
+ $count_ott_to_enst++;
}
}
- print "We have ".scalar(keys %enst_to_ott)." ott to enst entries\n " if($verbose);
+ print "We have $count_ott_to_enst ott to enst entries\n " if($verbose);
my $dbi = $self->dbi();
@@ -193,8 +195,7 @@ sub run_script {
my $xref_count = 0;
- foreach my $enst (keys %enst_to_ott){
- my $ott = $enst_to_ott{$enst};
+ foreach my $ott (keys %ott_to_enst){
if(defined($ott_to_vega_name{$ott})){
my $id = $curated_source_id;
my $name = $ott_to_vega_name{$ott};
@@ -211,10 +212,12 @@ sub run_script {
info_type => "DIRECT"} );
$xref_count++;
- $self->add_direct_xref($xref_id, $enst, "transcript", "");
+ foreach my $ensembl_stable_id (@{$ott_to_enst{$ott}}) {
+ $self->add_direct_xref($xref_id, $ensembl_stable_id, "transcript", "");
+ }
}
if(defined($ott_to_status{$ott})){
- $status_insert_sth->execute($enst, $ott_to_status{$ott});
+ $status_insert_sth->execute($ott_to_enst{$ott}, $ott_to_status{$ott});
}
}
@@ -247,4 +250,3 @@ sub run_script {
1;
-
--
GitLab