From 805432c6909e85f83845641d87ecd7c84c210b9e Mon Sep 17 00:00:00 2001
From: Magali Ruffier <mr6@ebi.ac.uk>
Date: Wed, 5 Sep 2018 16:09:25 +0100
Subject: [PATCH] ENSCORESW-2850: optimised for single species run
---
.../xref_mapping/XrefParser/ProcessData.pm | 49 +++++++++----------
misc-scripts/xref_mapping/xref_parser.pl | 14 ++----
2 files changed, 28 insertions(+), 35 deletions(-)
diff --git a/misc-scripts/xref_mapping/XrefParser/ProcessData.pm b/misc-scripts/xref_mapping/XrefParser/ProcessData.pm
index b38ad24eba..79a7fa50d1 100644
--- a/misc-scripts/xref_mapping/XrefParser/ProcessData.pm
+++ b/misc-scripts/xref_mapping/XrefParser/ProcessData.pm
@@ -55,7 +55,7 @@ sub run {
my $unzip = $ref_arg->{unzip};
my $stats = $ref_arg->{stats};
my $cleanup = $ref_arg->{cleanup};
- my $rspecies = $ref_arg->{speciesr};
+ my $species = $ref_arg->{species};
my $taxon_id = $ref_arg->{taxon};
my $division = $ref_arg->{division};
my $sources = $ref_arg->{sourcesr};
@@ -101,21 +101,23 @@ DSS
my $dep_sth = $dbi->prepare($sql);
# validate species names
- if (defined $division) { push @$rspecies, $division; }
- my @species_ids = $self->validate_species($rspecies, $verbose);
- if (defined $taxon_id) { push @species_ids, $taxon_id; }
+ my $species_id = $self->validate_species($species, $verbose);
+ my $division_id = $self->validate_species($division, $verbose);
+ my @species_sources = ($species_id);
+ push @species_sources, $division_id if defined $division_id;
+ push @species_sources, $taxon_id if defined $taxon_id;
# validate source names
- exit(1) if ( !$self->validate_sources(\@species_ids,$sources, $verbose) );
- exit(1) if ( !$self->validate_sources(\@species_ids,$notsources, $verbose) );
+ exit(1) if ( !$self->validate_sources(\@species_sources,$sources, $verbose) );
+ exit(1) if ( !$self->validate_sources(\@species_sources,$notsources, $verbose) );
# build SQL
my $species_sql = "";
- if (@species_ids) {
+ if (@species_sources) {
$species_sql .= " AND su.species_id IN (";
- for ( my $i = 0 ; $i < @species_ids ; $i++ ) {
+ for ( my $i = 0 ; $i < @species_sources; $i++ ) {
$species_sql .= "," if ( $i != 0 );
- $species_sql .= $species_ids[$i];
+ $species_sql .= $species_sources[$i];
}
$species_sql .= ") ";
}
@@ -156,12 +158,12 @@ DSS
$sth->execute();
my ( $source_id, $source_url_id, $name, $url, $release_url,
- $checksum, $parser, $species_id );
+ $checksum, $parser, $species_source_id );
$sth->bind_columns( \$source_id, \$source_url_id,
\$name, \$url,
\$release_url, \$checksum,
- \$parser, \$species_id );
+ \$parser, \$species_source_id );
my $dir;
my %summary = ();
@@ -241,7 +243,7 @@ DSS
foreach my $file (@files) {
# check dependencies are loaded all ready
- if(!($self->all_dependencies_loaded($source_id, $species_id, $name, $dep_sth))){
+ if(!($self->all_dependencies_loaded($source_id, $species_source_id, $name, $dep_sth))){
++$summary{$name}->{$parser};
next;
}
@@ -682,26 +684,21 @@ sub dbi {
###########################################################
sub validate_species {
my ($self, $species, $verbose) = @_;
- my @species_ids;
my $dbi = $self->dbi();
my $sth = $dbi->prepare("SELECT species_id, name FROM species WHERE LOWER(name)=? OR LOWER(aliases) REGEXP ?");
my ($species_id, $species_name);
- foreach my $sp (@$species) {
-
- my $bind_arg = "^".lc($sp).",|^".lc($sp)."\$|,[ ]{0,1}".lc($sp)."[ ]{0,1},|,[ ]{0,1}".lc($sp)."\$";
- $sth->execute(lc($sp), $bind_arg );
- $sth->bind_columns(\$species_id, \$species_name);
- if (my @row = $sth->fetchrow_array()) {
- print "Species $sp is valid (name = " . $species_name . ", ID = " . $species_id . ")\n" if($verbose);
- push @species_ids, $species_id;
- } else {
- print STDERR "Species $sp is not valid; valid species are:\n";
- $self->show_valid_species();
- }
+ my $bind_arg = "^".lc($species).",|^".lc($species)."\$|,[ ]{0,1}".lc($species)."[ ]{0,1},|,[ ]{0,1}".lc($species)."\$";
+ $sth->execute(lc($species), $bind_arg );
+ $sth->bind_columns(\$species_id, \$species_name);
+ if (my @row = $sth->fetchrow_array()) {
+ print "Species $species is valid (name = " . $species_name . ", ID = " . $species_id . ")\n" if($verbose);
+ } else {
+ print STDERR "Species $species is not valid; valid species are:\n";
+ $self->show_valid_species();
}
- return @species_ids;
+ return $species_id;
}
############################################################
diff --git a/misc-scripts/xref_mapping/xref_parser.pl b/misc-scripts/xref_mapping/xref_parser.pl
index 8f591ac512..a62052fff2 100644
--- a/misc-scripts/xref_mapping/xref_parser.pl
+++ b/misc-scripts/xref_mapping/xref_parser.pl
@@ -72,7 +72,6 @@ if($ARGV[0]){
exit(1);
}
-my @species = split(/,/,join(',',$species)) if $species;
my @sources = split(/,/,join(',',$sources)) if $sources;
my @notsource = split(/,/,join(',',$notsource)) if $notsource;
@@ -103,7 +102,7 @@ $process->run({ host => $host,
dbname => $dbname,
user => $user,
pass => $pass,
- speciesr => \@species,
+ species => $species,
taxon => $taxon,
division => $division,
sourcesr => \@sources,
@@ -147,7 +146,7 @@ sub usage {
print << "EOF";
xref_parser.pl -user {user} -pass {password} -host {host} \\
- -port {port} -dbname {database} -species {species1,species2} \\
+ -port {port} -dbname {database} -species {species} \\
-source {source1,source2} -notsource {source1,source2} \\
-create -setrelease -deletedownloaded -checkdownload -stats -verbose \\
-cleanup -drop_db -download_path -unzip
@@ -162,14 +161,11 @@ sub usage {
-dbname Name of xref database to use/create.
- -species Which species to import. Multiple -species arguments
- and/or comma, separated lists of species are
- allowed. Species may be referred to by genus/species
+ -species Which species to import.
+ Species may be referred to by genus/species
(e.g. homo_sapiens) or common aliases (e.g. human).
Specifying an unknown species will cause a list
- of valid species to be printed. Not specifying a
- -species argument will result in all species being
- used.
+ of valid species to be printed.
-source Which sources to import. Multiple -source arguments
and/or comma, separated lists of sources are
--
GitLab