[Bioperl-l] a problem with HspI module?

Chris Fields cjfields at illinois.edu
Fri May 21 02:24:46 UTC 2010


It would be best to file this in a bug report, along with example data.  

chris

On May 20, 2010, at 9:18 PM, Dimitar Kenanov wrote:

> Hello guys,
> i think i found a problem with ' Bio::Search::HSP::HSPI'. Consider the following HSP:
> -------------
> Score = 48.9 bits (115),  Expect = 8e-04, Method: Compositional matrix adjust.
> Identities = 27/77 (35%), Positives = 40/77 (51%), Gaps = 14/77 (18%)
> Frame = +1
> 
> Query  371      PSGMLLA-----SCSDDMTLKIWSMKQEVCIHDLQAHNKEIYTIKWSPTGPATSNPNSNI  425
>                P   LLA     S S D T+++W ++Q VC H L  H + +Y++ +SP G
> Sbjct  6955270  PGLQLLAFSHPPSASFDSTVRLWDVEQGVCTHTLMKHQEPVYSVAFSPDGK---------  6955422
> 
> Query  426      MLASASFDSTVRLWDIE  442
>                 LAS SFD  V +W+ +
> Sbjct  6955423  YLASGSFDKYVHIWNTQ  6955473
> ---------------
> 
> The method 'frac_identical' is not functioning right.
> -------------
> Title   : frac_identical
> Usage   : my $frac_id = $hsp->frac_identical( ['query'|'hit'|'total'] );
> Function: Returns the fraction of identitical positions for this HSP
> Returns : Float in range 0.0 -> 1.0
> Args    : 'query' = num identical / length of query seq (without gaps)
>           'hit'   = num identical / length of hit seq (without gaps)
>           'total' = num identical / length of alignment (with gaps)
>           default = 'total'
> ---------------
> According to the method description, for the HSP above, 'frac_identical' should return '0.42' with 'hit'. But it doesnt. Now with 'hit' gives '0.13'. With 'total' gives normal result '0.35'.
> 
> Thats all.
> Cheers
> 
> Dimitar
> 
> -- 
> Dimitar Kenanov
> Postdoctoral research fellow
> Protein Sequence Analysis Group
> Bioinformatics Institute
> A*STAR, Singapore
> tel: +65 6478 8514
> 
> _______________________________________________
> Bioperl-l mailing list
> Bioperl-l at lists.open-bio.org
> http://lists.open-bio.org/mailman/listinfo/bioperl-l





More information about the Bioperl-l mailing list