[Bioperl-l] how to "tile" the HSPs of a hit-object ?

Steve_Chervitz Steve_Chervitz at affymetrix.com
Tue Mar 21 14:15:50 EST 2006


Harald,

On Mar 21, 2006, at 7:40 AM, Harald wrote:

> Hi Steve.
>
> Thanks a lot for your help. But where is this contig-object  
> documented?

See the 'Returns' section of the method POD in my new version of  
SearchUtils.pm.
A contig is just a simple hash reference with 6 or so keys. I don't  
think we need a full-blown object for it yet.

Maybe we could provide access to it via the GenericHit object, so you  
don't have to call SearchUtils directly, and documentation would be  
easier to find.

> Besides, a small question about blast: Does the computation of the
> e-value (of a hit) take all HSPs into account or only a "best  
> alignment"?

The hit's e-value is for the top HSP, whose e-value could be based on  
either a single alignment or a group of alignments. If you see  
something like 'Expect(2)' in an NCBI blast report, this implies that  
sum statistics were applied to two HSPs to compute the expect score.  
Similarly for P(n) in WU-blast. This is only relevant for ungapped  
blast.

You can get the Expect(n) or P(n) number from the hit object or any  
HSP object calling the n() method. If you call n() on a hit from a  
gapped blast, you'll get the total number of HSPs.

Steve



More information about the Bioperl-l mailing list