From d322d934a653d7724ed3b0d94085fac0c854366d Mon Sep 17 00:00:00 2001
From: Matthew Laird <lairdm@ebi.ac.uk>
Date: Wed, 1 Jun 2016 16:35:36 +0100
Subject: [PATCH] Fix for ENSCORESW-1816, bioperl's inconsistent handling of
incomplete codons across versions. Trim any partial codons when translating
unless told not to.
---
modules/Bio/EnsEMBL/PredictionTranscript.pm | 29 ++++++++++++++++-----
modules/Bio/EnsEMBL/Transcript.pm | 19 ++++++++++++--
modules/t/transcript.t | 6 +++++
3 files changed, 45 insertions(+), 9 deletions(-)
diff --git a/modules/Bio/EnsEMBL/PredictionTranscript.pm b/modules/Bio/EnsEMBL/PredictionTranscript.pm
index 9ce0ab4b7d..041247253d 100644
--- a/modules/Bio/EnsEMBL/PredictionTranscript.pm
+++ b/modules/Bio/EnsEMBL/PredictionTranscript.pm
@@ -331,7 +331,8 @@ sub translation {
=head2 translate
- Args : none
+ Arg [1] : Boolean, emulate the behavior of old bioperl versions where
+ an incomplete final codon of 2 characters is padded and guessed
Function : Give a peptide translation of all exons currently in
the PT. Gives empty string when none is in.
Returntype: a Bio::Seq as in transcript->translate()
@@ -343,7 +344,7 @@ sub translation {
sub translate {
- my ($self) = @_;
+ my ($self, $complete_codon) = @_;
my $dna = $self->translateable_seq();
@@ -358,6 +359,11 @@ sub translate {
}
$codon_table_id ||= 1; #default will be vertebrates
+ # Remove the final stop codon from the mrna
+ # sequence produced if it is present, this is so any peptide produced
+ # won't have a terminal stop codon
+ # if you want to have a terminal stop codon either comment this line out
+ # or call translatable seq directly and produce a translation from it
if( CORE::length( $dna ) % 3 == 0 ) {
# $dna =~ s/TAG$|TGA$|TAA$//i;
my $codon_table = Bio::Tools::CodonTable->new( -id => $codon_table_id );
@@ -365,12 +371,21 @@ sub translate {
if ( $codon_table->is_ter_codon( substr( $dna, -3, 3 ) ) ) {
substr( $dna, -3, 3, '' );
}
+ } elsif ( CORE::length($dna) % 3 == 2 ) {
+ # If we have a partial codon of 2 bp we need to decide if we
+ # trim it or not to fix some bad behaviour in older bioperl
+ # versions
+ if ( $complete_codon ) {
+ # If we want to do the bad behavior of bioperl 1.6.1 and older
+ # where we guess the last codon if inomplete, pad an N
+ # to the mrna sequence
+ $dna .= 'N';
+ } else {
+ # Otherwise trim those last two bp off so the behavior is
+ # consistent across bioperl versions
+ substr( $dna, -2, 2, '' );
+ }
}
- # the above line will remove the final stop codon from the mrna
- # sequence produced if it is present, this is so any peptide produced
- # won't have a terminal stop codon
- # if you want to have a terminal stop codon either comment this line out
- # or call translatable seq directly and produce a translation from it
my $bioseq = new Bio::Seq( -id => $self->display_id,
-seq => $dna,
diff --git a/modules/Bio/EnsEMBL/Transcript.pm b/modules/Bio/EnsEMBL/Transcript.pm
index e41d0ffd81..9da16ac06b 100755
--- a/modules/Bio/EnsEMBL/Transcript.pm
+++ b/modules/Bio/EnsEMBL/Transcript.pm
@@ -2092,7 +2092,8 @@ sub get_all_translateable_Exons {
=head2 translate
- Args : none
+ Arg [1] : Boolean, emulate the behavior of old bioperl versions where
+ an incomplete final codon of 2 characters is padded and guessed
Example : none
Description: Return the peptide (plus eventual stop codon) for
this transcript. Does N-padding of non-phase
@@ -2107,7 +2108,7 @@ sub get_all_translateable_Exons {
=cut
sub translate {
- my ($self) = @_;
+ my ($self, $complete_codon) = @_;
if ( !defined( $self->translation() ) ) { return undef }
@@ -2152,6 +2153,20 @@ sub translate {
if ( $codon_table->is_ter_codon( substr( $mrna, -3, 3 ) ) ) {
substr( $mrna, -3, 3, '' );
}
+ } elsif ( CORE::length($mrna) % 3 == 2 ) {
+ # If we have a partial codon of 2 bp we need to decide if we
+ # trim it or not to fix some bad behaviour in older bioperl
+ # versions
+ if ( $complete_codon ) {
+ # If we want to do the bad behavior of bioperl 1.6.1 and older
+ # where we guess the last codon if inomplete, pad an N
+ # to the mrna sequence
+ $mrna .= 'N';
+ } else {
+ # Otherwise trim those last two bp off so the behavior is
+ # consistent across bioperl versions
+ substr( $mrna, -2, 2, '' );
+ }
}
if ( CORE::length($mrna) < 1 ) { return undef }
diff --git a/modules/t/transcript.t b/modules/t/transcript.t
index 77061d6a64..9b5cf356f3 100644
--- a/modules/t/transcript.t
+++ b/modules/t/transcript.t
@@ -210,6 +210,12 @@ $tr->flush_Exons();
is( scalar( @{$tr->get_all_Exons()} ), 0, 'No exons left after flushing' );
+# Fetch a fresh tr, check incomplete codon behavior
+$tr = $ta->fetch_by_stable_id( "ENST00000300425" );
+
+# By default the incomplete codon should be dropped
+is( $tr->translate()->seq() =~ /P$/, 1, "Incomplete codon is not translated");
+is( $tr->translate(1)->seq() =~ /PL$/, 1, "Incomplete codon is padded then translated");
# get a fresh tr to check the update method
$tr = $ta->fetch_by_stable_id( "ENST00000217347" );
--
GitLab