From ec9c73945f5028ea6c07744fe60ec73675805f4f Mon Sep 17 00:00:00 2001
From: Ian Longden <ianl@sanger.ac.uk>
Date: Wed, 6 Oct 2004 11:42:44 +0000
Subject: [PATCH] gen tidy
---
.../xref_mapping/XrefMapper/BasicMapper.pm | 173 ++++++++++++------
1 file changed, 119 insertions(+), 54 deletions(-)
diff --git a/misc-scripts/xref_mapping/XrefMapper/BasicMapper.pm b/misc-scripts/xref_mapping/XrefMapper/BasicMapper.pm
index 8e2950e762..5e105cc571 100644
--- a/misc-scripts/xref_mapping/XrefMapper/BasicMapper.pm
+++ b/misc-scripts/xref_mapping/XrefMapper/BasicMapper.pm
@@ -1,23 +1,47 @@
package XrefMapper::BasicMapper;
-
+
use strict;
use DBI;
use Bio::EnsEMBL::DBSQL::DBAdaptor;
use Bio::EnsEMBL::Translation;
+=head1 NAME
+
+XrefMapper::BasicMapper
+
+=head1 DESCIPTION
+
+This is the basic mapper routine. It will create the necessary fasta files for
+both the xref and ensembl sequences. These will then be matched using exonerate
+and the results written to another file. The xref database is hard coded at the
+beginning. By creating a <species>.pm file and inheriting from this base class
+different matching routines, parameters, data sets etc can be set.
+
+=head1 CONTACT
+
+Post questions to the EnsEMBL development list ensembl-dev@ebi.ac.uk
+
+
+=cut
+
+
+#
+# Specify xref database here
+#
my $xref_host = "ecs1g";
my $xref_port = 3306;
my $xref_database = "ianl_test_xref";
my $xref_user = "ensadmin";
my $xref_password = "ensembl";
-
-
+
+
+
sub new {
my($class, $species, $host, $port, $dbname, $user, $password ,$dir) = @_;
-
+
my $self ={};
bless $self,$class;
-
+
$self->species($species);
$self->host($host);
$self->port($port);
@@ -32,18 +56,26 @@ sub new {
-
sub dump_seqs{
my ($self, $slice) = @_;
$self->dump_ensembl($slice);
$self->dump_xref();
}
+sub run_matching{
+ print "NOT done yet:-)\n";
+}
+
+sub store{
+ print "NOT done yet Either :-)\n";
+}
sub dump_xref{
my ($self) = @_;
-
+ #
+ # the species specified must be in the database and hence have a species_id
+ #
my $sql = "select species_id from species where name = '".$self->species."'";
my $dbi = $self->dbi();
my $sth = $dbi->prepare($sql);
@@ -63,9 +95,13 @@ sub dump_xref{
}
die("Please try again\n");
}
-
- open(XDNA,">".$self->dir."/xref_dna.fasta") || die "Could not open xref_dna.fasta";
+ #
+ # Dump out the sequences where the species has an id of species_id
+ # and the sequence type is 'dna'
+ #
+ $self->xref_dna_file($self->dir."/".$self->species."_xref_dna.fasta");
+ open(XDNA,">".$self->dir."/".$self->species."_xref_dna.fasta") || die "Could not open xref_dna.fasta";
my $sql = "select p.xref_id, p.sequence from primary_xref p, xref x ";
$sql .= "where p.xref_id = x.xref_id and ";
$sql .= " p.sequence_type ='dna' and ";
@@ -84,7 +120,12 @@ sub dump_xref{
ENDDNA:
close XDNA;
- open(XPEP,">".$self->dir."/xref_prot.fasta") || die "Could not open xref_prot.fasta";
+ #
+ # Dump out the sequences where the species has an id of species_id
+ # and the sequence type is 'peptide'
+ #
+ $self->xref_protein_file($self->dir."/".$self->species."_xref_prot.fasta");
+ open(XPEP,">".$self->dir."/".$self->species."_xref_prot.fasta") || die "Could not open xref_prot.fasta";
$sql = "select p.xref_id, p.sequence from primary_xref p, xref x ";
$sql .= "where p.xref_id = x.xref_id and ";
$sql .= " p.sequence_type ='peptide' and ";
@@ -109,19 +150,14 @@ ENDXPEP:
sub dump_ensembl{
my ($self) = @_;
- #create filename
- $self->ensembl_protein_file($self->dir."/".$self->species."_protein.fasta");
- $self->ensembl_dna_file($self->dir."/".$self->species."_dna.fasta");
-
-
$self->fetch_and_dump_seq();
-
+
}
sub fetch_and_dump_seq{
- my ($self, $type, $adaptortype) = @_;
-
+ my ($self) = @_;
+
my $db = new Bio::EnsEMBL::DBSQL::DBAdaptor(-species => $self->species(),
-dbname => $self->dbname(),
-host => $self->host(),
@@ -129,10 +165,18 @@ sub fetch_and_dump_seq{
-password => $self->password(),
-user => $self->user(),
-group => 'core');
-
+
+ #
+ # store ensembl dna file name and open it
+ #
+ $self->ensembl_dna_file($self->dir."/".$self->species."_dna.fasta");
open(DNA,">".$self->ensembl_dna_file())
|| die("Could not open dna file for writing: ".$self->ensembl_dna_file."\n");
+ #
+ # store ensembl protein file name and open it
+ #
+ $self->ensembl_protein_file($self->dir."/".$self->species."_protein.fasta");
open(PEP,">".$self->ensembl_protein_file())
|| die("Could not open dna file for writing: ".$self->ensembl_protein_file."\n");
@@ -162,12 +206,45 @@ FIN:
close DNA;
close PEP;
}
-
+
+sub dbi {
+
+ my $self = shift;
+
+ my $dbi = DBI->connect("dbi:mysql:host=$xref_host;port=$xref_port;database=$xref_database",
+ "$xref_user",
+ "$xref_password",
+ {'RaiseError' => 1}) || die "Can't connect to database";
+
+
+ return $dbi;
+}
+
+###
+# Getter/Setter methods
+###
+
+sub xref_protein_file{
+ my ($self, $arg) = @_;
+
+ (defined $arg) &&
+ ($self->{_xref_prot_file} = $arg );
+ return $self->{_xref_prot_file};
+}
+
+sub xref_dna_file{
+ my ($self, $arg) = @_;
+
+ (defined $arg) &&
+ ($self->{_xref_dna_file} = $arg );
+ return $self->{_xref_dna_file};
+}
+
sub ensembl_protein_file{
my ($self, $arg) = @_;
-
+
(defined $arg) &&
($self->{_ens_prot_file} = $arg );
return $self->{_ens_prot_file};
@@ -175,42 +252,42 @@ sub ensembl_protein_file{
sub ensembl_dna_file{
my ($self, $arg) = @_;
-
+
(defined $arg) &&
($self->{_ens_dna_file} = $arg );
return $self->{_ens_dna_file};
}
-sub get_ensembl_type{
- my %type;
-
- $type{'Translation'} = "peptide";
- $type{'Transcript'} = "dna";
-
- return \%type;
-}
+#sub get_ensembl_type{
+# my %type;
+#
+# $type{'Translation'} = "peptide";
+# $type{'Transcript'} = "dna";
+#
+# return \%type;
+#}
sub species {
my ($self, $arg) = @_;
-
+
(defined $arg) &&
($self->{_species} = $arg );
return $self->{_species};
}
-
+
sub host {
my ($self, $arg) = @_;
-
+
(defined $arg) &&
($self->{_host} = $arg );
return $self->{_host};
}
-
+
sub port {
my ($self, $arg) = @_;
-
+
(defined $arg) &&
($self->{_port} = $arg );
return $self->{_port};
@@ -218,23 +295,23 @@ sub port {
sub dbname {
my ($self, $arg) = @_;
-
+
(defined $arg) &&
($self->{_dbname} = $arg );
return $self->{_dbname};
}
-
+
sub user {
my ($self, $arg) = @_;
-
+
(defined $arg) &&
($self->{_user} = $arg );
return $self->{_user};
}
-
+
sub password {
my ($self, $arg) = @_;
-
+
(defined $arg) &&
($self->{_password} = $arg );
return $self->{_password};
@@ -242,25 +319,13 @@ sub password {
sub dir {
my ($self, $arg) = @_;
-
+
(defined $arg) &&
($self->{_dir} = $arg );
return $self->{_dir};
}
-sub dbi {
-
- my $self = shift;
-
- my $dbi = DBI->connect("dbi:mysql:host=$xref_host;port=$xref_port;database=$xref_database",
- "$xref_user",
- "$xref_password",
- {'RaiseError' => 1}) || die "Can't connect to database";
-
-
- return $dbi;
-}
-
+
1;
--
GitLab