[Bioperl-l] 1 small help with WU-BLAST -postsw
jason.stajich at duke.edu
Sat Jan 21 10:38:23 EST 2006
if you are really using Hit->frac_aligned I wouldn't rely on the Hit
object frac_aligned - deal with the HSPs and calculate what you want
- if there are multiple HSPs after a postsw some of those could be
overalapping alternative sub-optimal alignments I don't know what
thethe Hit object algorithm does when it tries to merge these examples.
My best advice is
- find concrete examples of A,B and B,A pairs that doesn't meet
- print out the HSP information and LOOK at the HSPs and calculate
the expected numbers by hand to figure out what is going on
- there are going to be comparisons where it is ambiguous what
would you do BY HAND for these cases, ignore them because they are
the 5th best hit? take the longest HSP? try and
calculate overall coverage?
- calculate frac aligned per HSP, decide what you want to do when
there are multiple HSPs, take the longest, attempt to figure out some
overall coverage, it all depends on your
question if you are trying to find a single number
- $hit_HSP_frac_aligned = $hsp->hit->length / $hit->length
On Jan 20, 2006, at 4:34 PM, Angshu Kar wrote:
> But I'm doing an AND.So if i do use alignment/hit > 0.5 and
> alignment/query > 0.5 as a filter will the length of the denominator
> On 1/20/06, Jason Stajich <jason.stajich at duke.edu> wrote:
>> Well the hit and query probably are not the same length which is the
>> denominator in that fraction ....
>> On Jan 20, 2006, at 1:57 PM, Angshu Kar wrote:
>>> I'm using WU-BLASTP in yeast data (all vs all).
>>> Then I'm using :
>>> $hit_object->frac_aligned_query() > 0.5
>>> $hit_object->frac_aligned_hit() > 0.5
>>> as filter conditions.
>>> In that I'm getting asymmetric results! I mean I've sequences A,B in
>>> my o/p and not B,A. Has it got something to do with the asymmetry of
>>> BLAST (but I thought -postsw takes care of that)?
>>> Please help.
>>> Ignore the impossible but honor it ...
>>> The only enviable second position is success, since failure always
>>> comes first...
>>> Bioperl-l mailing list
>>> Bioperl-l at portal.open-bio.org
>> Jason Stajich
>> Duke University
> Ignore the impossible but honor it ...
> The only enviable second position is success, since failure always
> comes first...
More information about the Bioperl-l