From 506a9249117ca9f8587d90cde9533e1a2d7a3aba Mon Sep 17 00:00:00 2001
From: Arne Stabenau <stabenau@sanger.ac.uk>
Date: Thu, 21 Nov 2002 18:06:02 +0000
Subject: [PATCH] nt_contigs as slices supported
---
modules/Bio/EnsEMBL/DBSQL/SliceAdaptor.pm | 78 ++++++++++++++++++-----
modules/Bio/EnsEMBL/Slice.pm | 60 +++++++++++++++--
modules/t/slice.t | 14 +++-
modules/t/sliceAdaptor.t | 18 +++++-
4 files changed, 142 insertions(+), 28 deletions(-)
diff --git a/modules/Bio/EnsEMBL/DBSQL/SliceAdaptor.pm b/modules/Bio/EnsEMBL/DBSQL/SliceAdaptor.pm
index bada5f9fbf..5a07d51091 100644
--- a/modules/Bio/EnsEMBL/DBSQL/SliceAdaptor.pm
+++ b/modules/Bio/EnsEMBL/DBSQL/SliceAdaptor.pm
@@ -163,36 +163,80 @@ sub fetch_by_contig_name {
sub fetch_by_fpc_name {
my ($self,$fpc_name) = @_;
+
+ my( $p, $f, $l ) = caller;
+ $self->warn( "$f:$l calls deprecated method fetch_by_fpc_name. Please use fetch_by_supercontig_name instead" );
- my $type = $self->db->assembly_type();
+ $self->fetch_by_supercontig_name( $fpc_name );
+}
- my $sth = $self->db->prepare("
+sub fetch_by_supercontig_name {
+ my ($self,$supercontig_name) = @_;
+
+ my $assembly_type = $self->db->assembly_type();
+
+ my $sth = $self->db->prepare("
SELECT chr.name, a.superctg_ori, MIN(a.chr_start), MAX(a.chr_end)
FROM assembly a, chromosome chr
- WHERE superctg_name = '$fpc_name'
- AND type = '$type'
+ WHERE superctg_name = ?
+ AND type = ?
AND chr.chromosome_id = a.chromosome_id
GROUP by superctg_name
");
- $sth->execute;
+ $sth->execute( $supercontig_name, $assembly_type );
+
+ my ($chr, $strand, $slice_start, $slice_end) = $sth->fetchrow_array;
+
+ my $slice;
+
+ $slice = new Bio::EnsEMBL::Slice
+ (
+ -chr_name => $chr,
+ -chr_start =>$slice_start,
+ -chr_end => $slice_end,
+ -strand => $strand,
+ -assembly_type => $assembly_type
+ );
+
+ return $slice;
+}
- my ($chr, $strand, $slice_start, $slice_end) = $sth->fetchrow_array;
- my $slice;
+=head2 list_overlapping_supercontigs
- $slice = new Bio::EnsEMBL::Slice
- (
- -chr_name => $chr,
- -chr_start =>$slice_start,
- -chr_end => $slice_end,
- -strand => $strand,
- -assembly_type => $type
- );
+ Arg [1] : Bio::EnsEMBL::Slice $slice
+ overlapping given Sice
+ Example :
+ Description: return the names of the supercontigs that overlap given Slice.
+ Returntype : listref string
+ Exceptions : none
+ Caller : general
+
+=cut
- return $slice;
-}
+sub list_overlapping_supercontigs {
+ my ($self,$slice) = @_;
+ my $sth = $self->db->prepare( "
+ SELECT DISTINCT superctg_name
+ FROM assembly a, chromosome c
+ WHERE c.chromosome_id = a.chromosome_id
+ AND c.name = ?
+ AND a.type = ?
+ AND a.chr_end >= ?
+ AND a.chr_start <= ?
+ " );
+ $sth->execute( $slice->chr_name(), $slice->assembly_type(),
+ $slice->chr_start(), $slice->chr_end() );
+
+ my $result = [];
+ while( my $aref = $sth->fetchrow_arrayref() ) {
+ push( @$result, $aref->[0] );
+ }
+
+ return $result;
+}
=head2 fetch_by_clone_accession
diff --git a/modules/Bio/EnsEMBL/Slice.pm b/modules/Bio/EnsEMBL/Slice.pm
index 48690ac754..61204a9c71 100644
--- a/modules/Bio/EnsEMBL/Slice.pm
+++ b/modules/Bio/EnsEMBL/Slice.pm
@@ -192,7 +192,7 @@ sub dbID {
=head2 name
- Arg [1] : none
+ Arg [1] : optional string $name
Example : $name = $slice->name();
Description: Returns the name of this slice. The name is formatted as a
the following string: "$chr_name.$chr_start-$chr_end".
@@ -201,6 +201,7 @@ sub dbID {
can also act as a hash value. This is similar to the name
method in RawContig so for exons which can have either type
of sequence attached it provides a more common interface.
+ You can as well set the slicename to something like "NT_110023"
Returntype : string
Exceptions : none
Caller : general
@@ -208,18 +209,63 @@ sub dbID {
=cut
sub name {
- my $self = shift;
+ my ( $self, $arg ) = @_;
+
+ if( defined $arg ) {
+ $self->{name} = $arg;
+ } elsif(!defined $self->{name}) {
- my $string = join '', $self->chr_name, '.',
- $self->chr_start, '-', $self->chr_end();
+ my $string = join '', $self->chr_name, '.',
+ $self->chr_start, '-', $self->chr_end();
- if($self->strand == -1) {
- return "reverse($string)";
+
+ if($self->strand == -1) {
+ $self->{name} = "reverse($string)";
+ } else {
+ $self->{name} = $string;
+ }
}
- return $string;
+ return $self->{name};
}
+=head2 get_all_supercontig_Slices
+
+ Arg [1] : none
+ Example : none
+ Description: Returns Slices that represent overlapping supercontigs.
+ Coordinates inside those slices are supercontig coordinates.
+ You can transfer features to this slices coordinate system with
+ the normal transform call. The returned slices hav their names
+ set to the supercontig names.
+ Returntype : listref Bio::EnsEMBL::Slice
+ Exceptions : none
+ Caller : none
+
+=cut
+
+
+sub get_all_supercontig_Slices {
+ my $self = shift;
+ my $result = [];
+
+ if( $self->adaptor() ) {
+ my $superctg_names = $self->adaptor()->list_overlapping_supercontigs( $self );
+
+ for my $name ( @$superctg_names ) {
+ my $slice;
+ $slice = $self->adaptor()->fetch_by_supercontig_name( $name );
+ $slice->name( $name );
+ push( @$result, $slice );
+ }
+ } else {
+ $self->warn( "Slice needs to be attached to a database to get supercontigs" );
+ }
+
+ return $result;
+}
+
+
diff --git a/modules/t/slice.t b/modules/t/slice.t
index 21429fdb97..ddd6626761 100644
--- a/modules/t/slice.t
+++ b/modules/t/slice.t
@@ -5,13 +5,15 @@ use lib 't';
BEGIN { $| = 1;
use Test;
- plan tests => 46;
+ plan tests => 48;
}
+use TestUtils qw( debug );
use MultiTestDB;
use Bio::EnsEMBL::Slice;
+our $verbose= 0;
#
#1 TEST - Slice Compiles
@@ -270,6 +272,16 @@ $softmasked_seq = $seq = undef;
ok(scalar @{$slice->get_tiling_path});
+my $super_slices = $slice->get_all_supercontig_Slices();
+#
+# 47-48 get_all_supercontig_Slices()
+#
+debug( "Supercontig starts at ".$super_slices->[0]->chr_start() );
+
+ok( $super_slices->[0]->chr_start() == 29591966 );
+
+debug( "Supercontig name ".$super_slices->[0]->name() );
+ok( $super_slices->[0]->name() eq "NT_028392" );
diff --git a/modules/t/sliceAdaptor.t b/modules/t/sliceAdaptor.t
index 0d85ce760e..e1cfb80569 100644
--- a/modules/t/sliceAdaptor.t
+++ b/modules/t/sliceAdaptor.t
@@ -5,12 +5,14 @@ use warnings;
BEGIN { $| = 1;
use Test;
- plan tests => 26;
+ plan tests => 27;
}
use MultiTestDB;
use Bio::EnsEMBL::DBSQL::SliceAdaptor;
-use TestUtils qw(test_getter_setter);
+use TestUtils qw(test_getter_setter debug);
+
+our $verbose = 1;
my ($CHR, $START, $END, $FLANKING) = ("20", 30_252_000, 31_252_001, 1000);
@@ -68,10 +70,12 @@ ok($new_slice->chr_end == $slice->chr_end + $FLANKING);
# 12-13 fetch_by_fpc_name
#
my $fpc_name = 'NT_011387';
-$slice = $slice_adaptor->fetch_by_fpc_name($fpc_name);
+$slice = $slice_adaptor->fetch_by_supercontig_name($fpc_name);
ok($new_slice->chr_start);
ok($new_slice->chr_end);
+
+
#
# 14 - 15 fetch_by_clone_accession
#
@@ -145,4 +149,12 @@ my $chromo = $db->get_ChromosomeAdaptor->fetch_by_chr_name($CHR);
ok($chromo->length eq $slice->chr_end);
+$slice = $slice_adaptor->fetch_by_chr_start_end("20", 29_252_000, 31_252_001 );
+my $name_list = $slice_adaptor->list_overlapping_supercontigs( $slice );
+
+for my $name ( @$name_list ) {
+ debug( "Overlapping supercontig ".$name );
+}
+
+ok( $name_list->[0] eq "NT_028392" );
--
GitLab