[Bioperl-l] undefined sub-sequence with a single base

Sun Nov 30 21:07:58 UTC 2008

Steve,

If you can look through it that would be great, if it gets fixed for  
1.6 even better, but it's not a blocker by any means and I would  
rather have it fixed robustly vs. a problematic quickie fix 'for  
now'.  I think a documentation note indicating potential issues would  
suffice for me for the time being.  Sendu also made some changes on  
this for 1.5.2 if memory serves (made it a bit more robust), so it  
might be worth looking over the revision history.

Somewhat related: I noticed that Bio::Search::BlastUtils and  
Bio::Search::SearchUtils is redundant (both have the same methods,  
including tile_hsps, but the latter module seems to be more up-to- 
date).  Should we deprecate one for the other?  They are both used  
(but I'm sure SearchUtils could drop-in replace BlastUtils).

chris

On Nov 30, 2008, at 11:58 AM, Mark A. Jensen wrote:

> I think we're (Dave, Chris, and I) are  making progress on the  
> subsequence boundary issues. The fix should propagate through  
> everything having this issue...we hope--stay tuned...
> ----- Original Message ----- From: "Steve Chervitz" <sac at bioperl.org>
> To: "Jason Stajich" <jason at bioperl.org>
> Cc: "Chris Fields" <cjfields at illinois.edu>; <bioperl-l at lists.open-bio.org 
> >; "Alexie Papanicolaou" <apapanicolaou at ice.mpg.de>
> Sent: Sunday, November 30, 2008 12:50 PM
> Subject: Re: [Bioperl-l] undefined sub-sequence with a single base
>
>
>> Apologies for the trouble my tile_hsps method has caused. My original
>> intent for it was to provide approximate answers at the level of the
>> Hit object, providing a means of summing over all HSPs in a  
>> reasonable
>> way. I don't think it has ever gone through extensive testing with
>> different blast flavors and edge cases.
>>
>> What to do with it? Maybe provide warnings to it's limitations with
>> advice on other ways to proceed. Or bite the bullet and make it more
>> robust.
>>
>> As for the specific exception reported in this thread, it looks like
>> there's a bug causing it to not properly handle zero-length ranges:
>>
>> "Undefined sub-sequence (548,548). Valid range = 51 - 548"
>>
>> Patching this might be fairly straightforward and could help in the
>> short term. I'll take a look.
>>
>> Steve
>>
>>
>> On Mon, Nov 17, 2008 at 10:16 AM, Jason Stajich <jason at bioperl.org>  
>> wrote:
>>> Personally - I'm not sure I trust tile_hsps on a translated search  
>>> - or at
>>> all - really - you may want to compute the "dominant" strand  
>>> yourself by
>>> iterating through the HSPs or using WU-BLAST to get logical groups  
>>> of HSPs
>>> which is a better tiling HSP algorithm (the --links option in WU- 
>>> BLAST).
>>>
>>> -jason
>>> On Nov 17, 2008, at 9:43 AM, Alexie Papanicolaou wrote:
>>>
>>>> Hi Chris
>>>>
>>>> Sorry, I got the new SVN build today and still get the same  
>>>> error...
>>>>
>>>> Could it be because the subseq is not divisible by 3 (due to  
>>>> blastx)?
>>>>
>>>> a
>>>>
>>>>
>>>> ------------- EXCEPTION: Bio::Root::Exception -------------
>>>> MSG: Undefined sub-sequence (2,2). Valid range = 2 - 190
>>>> STACK: Error::throw
>>>> STACK:
>>>> Bio::Root::Root::throw /usr/local/share/perl/5.8.8/Bio/Root/ 
>>>> Root.pm:357
>>>> STACK:
>>>> Bio::Search::HSP::HSPI::matches
>>>> /usr/local/share/perl/5.8.8/Bio/Search/HSP/HSPI.pm:691
>>>> STACK:
>>>> Bio::Search::SearchUtils::_adjust_contigs
>>>> /usr/local/share/perl/5.8.8/Bio/Search/SearchUtils.pm:460
>>>> STACK:
>>>> Bio::Search::SearchUtils::tile_hsps
>>>> /usr/local/share/perl/5.8.8/Bio/Search/SearchUtils.pm:200
>>>> STACK:
>>>> Bio::Search::Hit::GenericHit::strand
>>>> /usr/local/share/perl/5.8.8/Bio/Search/Hit/GenericHit.pm:1455
>>>>
>>>>
>>>> ------------- EXCEPTION: Bio::Root::Exception -------------
>>>> MSG: Undefined sub-sequence (3,4). Valid range = 3 - 44
>>>> STACK: Error::throw
>>>> STACK:
>>>> Bio::Root::Root::throw /usr/local/share/perl/5.8.8/Bio/Root/ 
>>>> Root.pm:357
>>>> STACK:
>>>> Bio::Search::HSP::HSPI::matches
>>>> /usr/local/share/perl/5.8.8/Bio/Search/HSP/HSPI.pm:691
>>>> STACK:
>>>> Bio::Search::SearchUtils::_adjust_contigs
>>>> /usr/local/share/perl/5.8.8/Bio/Search/SearchUtils.pm:404
>>>> STACK:
>>>> Bio::Search::SearchUtils::tile_hsps
>>>> /usr/local/share/perl/5.8.8/Bio/Search/SearchUtils.pm:200
>>>> STACK:
>>>> Bio::Search::Hit::GenericHit::strand
>>>> /usr/local/share/perl/5.8.8/Bio/Search/Hit/GenericHit.pm:1455
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> On Fri, 2008-11-14 at 11:08 -0600, Chris Fields wrote:
>>>>
>>>>> We've switched to subversion a while ago.  Could you try  
>>>>> updating from
>>>>> there, or using one of our nightly builds?
>>>>>
>>>>> http://www.bioperl.org/DIST/nightly_builds/
>>>>>
>>>>> chris
>>>>>
>>>>> On Nov 14, 2008, at 8:59 AM, Alexie Papanicolaou wrote:
>>>>
>>>>
>>>> _______________________________________________
>>>> Bioperl-l mailing list
>>>> Bioperl-l at lists.open-bio.org
>>>> http://lists.open-bio.org/mailman/listinfo/bioperl-l
>>>
>>> Jason Stajich
>>> jason at bioperl.org
>>>
>>>
>>>
>>> _______________________________________________
>>> 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
>>
>
> _______________________________________________
> Bioperl-l mailing list
> Bioperl-l at lists.open-bio.org
> http://lists.open-bio.org/mailman/listinfo/bioperl-l