From cc1793300c1a53bc56b65efe6c6b0e2d658f0dea Mon Sep 17 00:00:00 2001
From: Glenn Proctor <gp1@sanger.ac.uk>
Date: Mon, 29 Jan 2007 10:07:56 +0000
Subject: [PATCH] Added dumping of variation IDs (on by default, can be
switched off with -no_variation option)
---
misc-scripts/ebi_search_dump/dump_ebi.pl | 82 ++++++++++++++++++++----
1 file changed, 69 insertions(+), 13 deletions(-)
diff --git a/misc-scripts/ebi_search_dump/dump_ebi.pl b/misc-scripts/ebi_search_dump/dump_ebi.pl
index 5a826f67d9..e3879eafc6 100644
--- a/misc-scripts/ebi_search_dump/dump_ebi.pl
+++ b/misc-scripts/ebi_search_dump/dump_ebi.pl
@@ -10,11 +10,13 @@ use DBI;
use Getopt::Long;
use IO::Zlib;
+use Bio::EnsEMBL::Registry;
use Bio::EnsEMBL::DBSQL::DBAdaptor;
+use Bio::EnsEMBL::Variation::DBSQL::DBAdaptor;
use HTML::Entities;
-my ( $host, $user, $pass, $port, $dbpattern, $max_genes, $gzip );
+my ($host, $user, $pass, $port, $dbpattern, $max_genes, $gzip, $no_variation);
GetOptions( "host=s", \$host,
"user=s", \$user,
@@ -23,6 +25,7 @@ GetOptions( "host=s", \$host,
"dbpattern|pattern=s", \$dbpattern,
"gzip!", \$gzip,
"max_genes=i", \$max_genes,
+ "no_variation", \$no_variation,
"help" , \&usage
);
@@ -38,6 +41,11 @@ run();
sub run() {
+ Bio::EnsEMBL::Registry->load_registry_from_db(-host => $host,
+ -port => $port,
+ -user => $user,
+ -pass => $pass);
+
# loop over databases
my $dsn = "DBI:mysql:host=$host";
@@ -130,6 +138,13 @@ sub content {
my $meta_container = $dba->get_MetaContainer();
my $species = $meta_container->get_Species()->common_name();
+ my $db_variation = variation_attach($dba) unless $no_variation;
+
+ my $trv_adaptor;
+ if ($db_variation) { # not all species have variation databases
+ $trv_adaptor = $db_variation->get_TranscriptVariationAdaptor();
+ }
+
foreach my $gene (@{$gene_adaptor->fetch_all()}) {
last if ($max_genes && $entry_count >= $max_genes);
@@ -175,7 +190,8 @@ sub content {
# additional fields - transcript, translation, species etc
p ("<additional_fields>");
- foreach my $transcript (@{$gene->get_all_Transcripts()}) {
+ my $transcripts = $gene->get_all_Transcripts();
+ foreach my $transcript (@{$transcripts}) {
p ("<field name=\"transcript\">" . $transcript->stable_id() . "</field>");
@@ -190,6 +206,13 @@ sub content {
p ("<field name=\"species\">" . $species . "</field>");
+ # SNP IDs
+ if ($db_variation) {
+ foreach my $tv (@{$trv_adaptor->fetch_all_by_Transcripts($transcripts)}){
+ p ("<field name=\"variation_id\">" . $tv->variation_feature()->variation_name() . "</field>");
+ }
+ }
+
p ("</additional_fields>");
# close tag
@@ -199,8 +222,6 @@ sub content {
}
-
-
}
# -------------------------------------------------------------------------------
@@ -223,7 +244,7 @@ sub footer {
} else {
close(FILE);
}
-
+
}
@@ -309,26 +330,61 @@ sub print_time {
# -------------------------------------------------------------------------------
+#
+# Figure out the name of a variation database from the core database name
+#
+
+sub variation_attach {
+
+ my $db = shift;
+
+ my $core_db_name;
+ $core_db_name = $db->dbc->dbname();
+ return undef if ($core_db_name !~ /_core_/);
+
+ my $dbc = $db->dbc();
+ my $sth = $dbc->prepare("show databases");
+ $sth->execute();
+ my $all_db_names = $sth->fetchall_arrayref();
+ my %all_db_names = map {( $_->[0] , 1)} @$all_db_names;
+ my $variation_db_name = $core_db_name;
+ $variation_db_name =~ s/_core_/_variation_/;
+
+ return undef if (! exists $all_db_names{$variation_db_name});
+
+ # register the dbadaptor with the Registry
+ return Bio::EnsEMBL::Variation::DBSQL::DBAdaptor->new(-host => $dbc->host(),
+ -user => $dbc->username(),
+ -pass => $dbc->password(),
+ -port => $dbc->port(),
+ -dbname => $variation_db_name);
+
+}
+
+# -------------------------------------------------------------------------------
+
sub usage {
print <<EOF; exit(0);
Usage: perl $0 <options>
- -host Database host to connect to.
+ -host Database host to connect to.
+
+ -port Database port to connect to.
- -port Database port to connect to.
+ -dbpattern Database name regexp
- -dbpattern Database name regexp
+ -user Database username.
- -user Database username.
+ -pass Password for user.
- -pass Password for user.
+ -gzip Compress output as it's written.
- -gzip Compress output as it's written.
+ -max_genes Only dump this many genes for testing.
- -max_genes Only dump this many genes for testing.
+ -no_variation Don't dump variation IDs.
- -help This message.
+ -help This message.
EOF
--
GitLab