From cbeca63af15421ada9820bf3a1435172f92d112b Mon Sep 17 00:00:00 2001
From: Ian Longden <ianl@sanger.ac.uk>
Date: Wed, 17 Nov 2004 11:18:19 +0000
Subject: [PATCH] check the species is correct. This will make the parsing a
little slower for those species specific files but the same modules can now
be used if you have to parse a file with a mixture of species.
---
.../xref_mapping/XrefParser/GOParser.pm | 48 ++++++++++---------
.../XrefParser/RefSeqGPFFParser.pm | 22 +++++----
.../xref_mapping/XrefParser/RefSeqParser.pm | 20 +++++---
3 files changed, 53 insertions(+), 37 deletions(-)
diff --git a/misc-scripts/xref_mapping/XrefParser/GOParser.pm b/misc-scripts/xref_mapping/XrefParser/GOParser.pm
index 1739761ae5..1874f84698 100644
--- a/misc-scripts/xref_mapping/XrefParser/GOParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/GOParser.pm
@@ -47,31 +47,35 @@ sub run {
open(GO,"<".$file) || die "Could not open $file\n";
+ my $taxon_line = "taxon:".$species_id;
+
while (<GO>) {
- chomp;
- my @array = split (/\t/,$_);
- $array[9] =~ s/\'/\\\'/g;
- my $master=0;
- if($array[0] =~ /ENSEMBL/){
- #these might be good for a check
- # match GO to Uniprot
- # match Uniprot to ENSEMBL
- # check ENSEMBL's are the same.
- }
- elsif($array[0] =~ /RefSeq/){
- if($refseq{$array[1]}){
- XrefParser::BaseParser->add_to_xrefs($refseq{$array[1]},$array[4],'',$array[4],'',$array[6],$source_id,$species_id);
- $count++;
+ if(/$taxon_line/){
+ chomp;
+ my @array = split (/\t/,$_);
+ $array[9] =~ s/\'/\\\'/g;
+ my $master=0;
+ if($array[0] =~ /ENSEMBL/){
+ #these might be good for a check
+ # match GO to Uniprot
+ # match Uniprot to ENSEMBL
+ # check ENSEMBL's are the same.
}
- }
- elsif($array[0] =~ /UniProt/){
- if($swiss{$array[1]}){
- XrefParser::BaseParser->add_to_xrefs($swiss{$array[1]},$array[4],'',$array[4],'',$array[6],$source_id,$species_id);
- $count++;
+ elsif($array[0] =~ /RefSeq/){
+ if($refseq{$array[1]}){
+ XrefParser::BaseParser->add_to_xrefs($refseq{$array[1]},$array[4],'',$array[4],'',$array[6],$source_id,$species_id);
+ $count++;
+ }
+ }
+ elsif($array[0] =~ /UniProt/){
+ if($swiss{$array[1]}){
+ XrefParser::BaseParser->add_to_xrefs($swiss{$array[1]},$array[4],'',$array[4],'',$array[6],$source_id,$species_id);
+ $count++;
+ }
+ }
+ else{
+ print STDERR "unknown type ".$array[0]."\n";
}
- }
- else{
- print STDERR "unknown type ".$array[0]."\n";
}
}
print "\t$count GO dependent xrefs added\n";
diff --git a/misc-scripts/xref_mapping/XrefParser/RefSeqGPFFParser.pm b/misc-scripts/xref_mapping/XrefParser/RefSeqGPFFParser.pm
index 0d4f0685a9..a2db83927d 100644
--- a/misc-scripts/xref_mapping/XrefParser/RefSeqGPFFParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/RefSeqGPFFParser.pm
@@ -17,7 +17,7 @@ use vars qw(@ISA);
if (!defined(caller())) {
if (scalar(@ARGV) != 1) {
- print "\nUsage: RefSeqGPFFParser.pm file.SPC\n\n";
+ print "\nUsage: RefSeqGPFFParser.pm file.SPC <source_id>\n\n";
exit(1);
}
@@ -32,13 +32,16 @@ sub run {
my $self = shift if (defined(caller(1)));
my $file = shift;
my $source_id = shift;
+ my $species_id = shift;
if ($source_id < 1) {
$source_id = XrefParser::BaseParser->get_source_id_for_filename(basename($file));
- print "Source id for $file: $source_id\n";
+ }
+ if(!defined($species_id)){
+ $species_id = XrefParser::BaseParser->get_species_id_for_filename($file);
}
- XrefParser::BaseParser->upload_xrefs(create_xrefs($source_id, $file));
+ XrefParser::BaseParser->upload_xrefs(create_xrefs($source_id, $file, $species_id));
}
@@ -51,7 +54,7 @@ sub run {
sub create_xrefs {
- my ($source_id, $file) = @_;
+ my ($source_id, $file, $species_id) = @_;
my %name2species_id = XrefParser::BaseParser->name2species_id();
@@ -76,10 +79,10 @@ sub create_xrefs {
$species =~ s/^\s*//g;
$species =~ s/\s+/_/g;
$species =~ s/\n//g;
- my $species_id = $name2species_id{$species};
-
+ my $species_id_check = $name2species_id{$species};
+
# skip xrefs for species that aren't in the species table
- if (defined $species_id) {
+ if (defined ($species_id) and $species_id = $species_id_check) {
my ($acc) = $entry =~ /ACCESSION\s+(\S+)/;
my ($ver) = $entry =~ /VERSION\s+(\S+)/;
@@ -163,7 +166,10 @@ sub create_xrefs {
}
push @xrefs, $xref;
- } # if defined species
+ }# if defined species
+ else{ #### REMOVE after TESTING
+ print "not correct $species $species_id NE $species_id_check\n";
+ }
} # while <REFSEQ>
diff --git a/misc-scripts/xref_mapping/XrefParser/RefSeqParser.pm b/misc-scripts/xref_mapping/XrefParser/RefSeqParser.pm
index a617465349..aa41c15743 100644
--- a/misc-scripts/xref_mapping/XrefParser/RefSeqParser.pm
+++ b/misc-scripts/xref_mapping/XrefParser/RefSeqParser.pm
@@ -17,7 +17,7 @@ use vars qw(@ISA);
if (!defined(caller())) {
if (scalar(@ARGV) != 1) {
- print "\nUsage: RefSeqParser.pm file.SPC\n\n";
+ print "\nUsage: RefSeqParser.pm file.SPC <source_id> <species_id>\n\n";
exit(1);
}
@@ -32,13 +32,16 @@ sub run {
my $self = shift if (defined(caller(1)));
my $file = shift;
my $source_id = shift;
+ my $species_id = shift;
- if ($source_id < 1) {
+ if (!defined($source_id) or $source_id < 1) {
$source_id = XrefParser::BaseParser->get_source_id_for_filename(basename($file));
- print "Source id for $file: $source_id\n";
+ }
+ if(!defined($species_id)){
+ $species_id = XrefParser::BaseParser->get_species_id_for_filename($file);
}
- XrefParser::BaseParser->upload_xrefs(create_xrefs($source_id, $file));
+ XrefParser::BaseParser->upload_xrefs(create_xrefs($source_id, $file, $species_id));
}
@@ -51,7 +54,7 @@ sub run {
sub create_xrefs {
- my ($source_id, $file) = @_;
+ my ($source_id, $file, $species_id) = @_;
my %name2species_id = XrefParser::BaseParser->name2species_id();
@@ -92,10 +95,10 @@ sub create_xrefs {
$species = lc $species;
$species =~ s/ /_/;
- my $species_id = $name2species_id{$species};
+ my $species_id_check = $name2species_id{$species};
# skip xrefs for species that aren't in the species table
- if (defined $species_id) {
+ if (defined($species_id) and $species_id = $species_id_check) {
my ($acc_no_ver,$ver) = split (/\./,$acc);
$xref->{ACCESSION} = $acc_no_ver;
@@ -111,6 +114,9 @@ sub create_xrefs {
push @xrefs, $xref;
}
+ else{
+ print "not correct species $species_id NE $species_id_check\n";
+ }
}
--
GitLab