[Biopython] NCBIWWW.qblast Expect value / E-value

[Peter Cock]

On Wed, May 20, 2009 at 1:51 PM, Jasper Koehorst
<jasperkoehorst at gmail.com> wrote:
> Im running against problems when i try to identify short peptide sequences
> of ±10 - 20 peptides. When i run this at the NCBI Blast website i get
> results, these all have
> high E-Values that i dont really care at the moment. The problem is when i
> do this in biopython as stated below, i will not get any results...
>
> I believe this is due to the fact that biopython will not show results with
> a "high" E-Value. Is there a way to change this? So it will allow results
> with an E-value of ±1500 or more?

Nothing in Biopython limits the expectation values - our qblast function
defaults to 10, but you can set this to what you like. However, the NCBI
may be imposing their own limit. Are you sure using anything more than
10 is actually meaningful?

> I tried the sript below but that does not quit work...
>
> Anybody has an idea?
>
>        result_handle =
> NCBIWWW.qblast("blastp","nr",peptide[1],entrez_query=i, expect=2000,
> matrix_name='BLOSUM80')
>        blast_records = NCBIXML.parse(result_handle)
>        blast_record = blast_records.next()
>        E_VALUE_TRESH = 10000000
>        for alignment in blast_record.alignments:
>            for hsp in alignment.hsps:
>

I would guess (from previous examples) that this is due to the NCBI
website and QBLAST API using different default parameters - the
NCBI likes to change the defaults on the website from time to time,
and these may differ from what you are getting via their QBLAST
API. I would start by checking the gap parameters.

See also:
http://lists.open-bio.org/pipermail/biopython/2008-May/004252.html
http://lists.open-bio.org/pipermail/biopython/2007-August/003679.html

Peter