[Bioperl-l] Bio::Tools::Glimmer and genes wrapping around the origin

[Fields, Christopher J]

The best way to address this is to propose a set of tests that demonstrates the problem, but I believe you have something that should work:

    my $location_string = 'join(117..1,135690..135187)';

We probably should cover both aspects, a join that spans the origin on both forward and reverse strands, something like:

   join(117..1,135690..135187)
   complement(join(117..1,135690..135187))

chris

On Mar 9, 2012, at 5:02 AM, Adam Witney wrote:

> 
> Thanks for your replies.
> 
> After a little more digging there seems to be two things. In the test code i posted, the extra strand attribute in Bio::SeqFeature::Generic (and subsequent call to location_object->strand) changes the internal guide_strand such that this line in Bio::Location::Split->to_FTstring changes the order of subLocations:
> 
> 	my @locs = ($stype eq 'join' && (!$guide && $strand == -1)) ?
> 	           reverse $self->sub_Location() : $self->sub_Location() ;
> 
> The second problem is probably due to the sorting order within Bio::SeqFeature::Generic when calling 'start' and 'end' then it doesn't do the right thing in this case. But I am not quite sure how to fix this.
> 
> Thanks again
> 
> Adam
> 
> On 8 Mar 2012, at 21:38, Fields, Christopher J wrote:
> 
>> I agree, no sorting should be implied (a 'join' order should be based on order of addition alone, and sort should be optional).  IIRC there were backwards-compat problems switching this due to reliance on old behavior, but it might be worth trying to see if anything breaks test-wise (and why it breaks).
>> 
>> chris
>> 
>> On Mar 8, 2012, at 2:37 PM, Francisco J. Ossandón wrote:
>> 
>>> Long ago, Bio::SeqIO had an issue with genes that were split at the origin.
>>> This was because Bioperl, when reading a Genbank file, automatically sorted
>>> the segments coordinates of the genes that was reading (not considering the
>>> possibility in a circular genome, where the sequence could go from the 1st
>>> nucleotide directly to the last one), so an extra "-nosort" argument was
>>> necessary every time to avoid Bioperl giving the wrong sequence:
>>> 
>>> my $ nt_seq_obj = $feat->spliced_seq(-nosort => 1);
>>> 
>>> This is the same bug. In that case the code was changed so the "-nosort"
>>> were applied based on the status of "is_circular" of the genome, see here:
>>> https://redmine.open-bio.org/issues/2579
>>> 
>>> Since your code don't have the "is_circular" information (because it don't
>>> come from a file), I guess that the autosorting is kicking in. I think it
>>> would be better if all the "autosorting" of the sublocations array inside
>>> "Bio::Location::Split" were optional instead of automatic, because of these
>>> cases.
>>> 
>>> Cheers,
>>> 
>>> Francisco J. Ossandon
>>> 
>>> -----Mensaje original-----
>>> De: bioperl-l-bounces at lists.open-bio.org
>>> [mailto:bioperl-l-bounces at lists.open-bio.org] En nombre de Adam Witney
>>> Enviado el: jueves, 08 de marzo de 2012 13:40
>>> Para: bioperl-l at bioperl.org
>>> Asunto: [Bioperl-l] Bio::Tools::Glimmer and genes wrapping around the origin
>>> 
>>> Hi,
>>> 
>>> I have been using Bio::Tools::Glimmer and have come across a problem with it
>>> not handling genes that wraparound across the origin. I think I have boiled
>>> it down to this test case of what happens internally with
>>> Bio::Tools::Glimmer
>>> 
>>> ##############################################################
>>> #! /usr/local/bin/perl -w
>>> 
>>> use strict;
>>> use warnings;
>>> 
>>> use Bio::Factory::FTLocationFactory;
>>> use Bio::SeqFeature::Generic;
>>> 
>>> my $location_string = 'join(117..1,135690..135187)';
>>> 
>>> my $location_factory = Bio::Factory::FTLocationFactory->new();
>>> my $location_object = $location_factory->from_string($location_string);
>>> 
>>> print "Location: ".$location_object->to_FTstring."\n";
>>> 
>>> my $gene = Bio::SeqFeature::Generic->new(
>>>               '-seq_id'      => 'Testing',
>>>               '-location'   => $location_object,
>>>               '-strand'     => -1
>>>           );
>>> 
>>> print "Location: ".$location_object->to_FTstring."\n";
>>> 
>>> ##############################################################
>>> 
>>> $ perl ../FTLocationTest.pl 
>>> Location: complement(join(135187..135690,1..117))
>>> Location: complement(join(1..117,135187..135690))
>>> 
>>> This happens because by setting the '-strand' in Bio::SeqFeature::Generic,
>>> this calls the strand method in $location_object (Bio::Location::Split)
>>> which then causes the problem (although I can't quite work out where...!
>>> 
>>> Is this intended behaviour?
>>> 
>>> Thanks
>>> 
>>> Adam
>>> _______________________________________________
>>> Bioperl-l mailing list
>>> Bioperl-l at lists.open-bio.org
>>> http://lists.open-bio.org/mailman/listinfo/bioperl-l
>>> 
>>> 
>>> _______________________________________________
>>> Bioperl-l mailing list
>>> Bioperl-l at lists.open-bio.org
>>> http://lists.open-bio.org/mailman/listinfo/bioperl-l
>> 
>