[Bioperl-l] bioperl script error

Andreo Beck andreo_beck at yahoo.com
Wed Jan 4 20:40:32 EST 2006


I've used both WU-BLASTP with -postsw as well as FASTA (WU-BLAST doesn't take other formats without transformations), And there is a sad part with the -postsw is that it doesn't work with BLASTN.
Anyways, now there are 2 questions that I've:
1. _adjust_contigs documentation says that its in experimental stage. What future enhancements are planned? Any plans to handle gapped alignments?
2. is the tile_hsps calculating the total length of the alignment? Then is that the point where things are breaking?



Jason Stajich <jason.stajich at duke.edu> wrote: 
On Jan 4, 2006, at 8:13 PM, Andreo Beck wrote:

My line 30 has:

$hit->hsp->evalue <= $threshold

But strangely enough, when I ran it the last time it ran successfully.
One point is that, in my line 32 I've added:

� �� if(($hit->length_aln() >= $sum_of_HSP_len && $hit->length()>= $hit_len ))


Now its clear that this is the line that's creating the problems, more specifically the length_aln function! The calls as you see are length_aln --> tile_hsps --> _adjust_contigs-->matches. Can this have anything to do with overlaps in my data?


Absolutely.� That is the function causing you problems, as to why it is failing, I am guessing it can't figure out how to merge the HSPs.� Personally I leave this sort of thing to the alignment rather than post-processing what BLAST did.� Depending on your question/compute/time you can get a better HSP path by running�
* WU-BLAST with the -postsw option which will cleanup these overlapping alignments
* FASTA or�SSEARCH to get a single alignment path.

 

Jason Stajich <jason.stajich at duke.edu> wrote: What does line 30 in your script say?
>> STACK toplevel /var/spool/slurmd/job01852/script:30

It is trying to merge HSPs to compute a the virtual length of a hit  
from the sub-HS! Ps.  This happens when you call a function on the HitI  
object like start/end/strand that has to merge the HSPs in order to  
get some sort of overall start/end/length for an alignment.  I don't  
really know what is going on in those Hit functions but they  
shouldn't crash if the HSP path set is unresolveable.  (I bet you are  
thinking the same thing...)  Are you sure you want to be calling  
whatever function it is on line 30?

-jason
On Jan 4, 2006, at 6:58 PM, Andreo Beck wrote:

> Great Torsten. But I wonder where this function is being called.  
> Also even if it is called when my code is supplying out of range  
> coordinates?
> I'm using BLASTP.
>
> Torsten Seemann  wrote: Andy,
>
>> Any clue what may be the cause of this exception? The sub-sequence  
>> seems to be within the valid range.
>>  ------------- EXCEPTION  -------------
&! gt;> MSG: Undefined sub-sequence (75,77). Valid range = 17 - 77
>> STACK Bio::Search::HSP::HSPI::matches /cmpchome/andyb/lib/perl/Bio/ 
>> Search/HSP/HSPI.pm:711
>> STACK (eval) /cmpchome/andyb/lib/perl/Bio/Search/SearchUtils.pm:365
>> STACK Bio::Search::SearchUtils::_adjust_contigs /cmpchome/andyb/ 
>> lib/perl/Bio/Search/SearchUtils.pm:364
>> STACK Bio::Search::SearchUtils::tile_hsps /cmpchome/andyb/lib/perl/ 
>> Bio/Search/SearchUtils.pm:176
>> STACK Bio::Search::Hit::GenericHit::length_aln /cmpchome/andyb/lib/ 
>> perl/Bio/Search/Hit/GenericHit.pm:740
>> STACK toplevel /var/spool/slurmd/job01852/script:30
>
> What type of report are you processing?
> eg. [T]BLAST[NX], BLASTP, HMMER
>
> Here's the problem causing function in bioperl-live:
> http://doc.bioperl.org/bioperl-live/Bio/Search/HSP/HSPI.html#CODE22
>
> -- 
> Torsten Seemann> Victorian Bioinformatics Consortium, Monash University, Australia
> http://www.vicbioinformatics.com/
>
>
>
>   
> ---------------------------------
>  Yahoo! DSL Something to write home about. Just $16.99/mo. or less
> _______________________________________________
> Bioperl-l mailing list
> Bioperl-l at portal.open-bio.org
> http://portal.open-bio.org/mailman/listinfo/bioperl-l

--
Jason Stajich
Duke University
http://www.duke.edu/~jes12


_______________________________________________
Bioperl-l mailing list
Bioperl-l at portal.open-bio.org
http://portal.open-bio.org/mailman/listinfo/bioperl-l


   


---------------------------------
 Yahoo! DSL Something to write home about. Just $16.99/mo. or less

 --
Jason Stajich
Duke University
http://www.duke.edu/~jes12

 




		
---------------------------------
 Yahoo! DSL Something to write home about. Just $16.99/mo. or less


More information about the Bioperl-l mailing list