[Biopython] qblast results (NCBIWWW module) different from web blast results

[Brad Chapman]

Alon;

> when I run a simple remote blast search using qblast of the biopython  
> package i get different results from some sequences from what I would  
> get for the same sequences when I preform
> the search on the NCBI Blast web site. specifically, while the hit  
> seems to be the same (same organism, protein and frame) the evalue is  
> significantly  lower with the qblast.

Could you provide more details about the query sequence and
database? I was not able to replicate this with a test sequence
against the nr database. qblast and the web interface are
using the same versions and have the same number of database
sequences.

> needless to say I have used the same parameters for both searches.
> interestingly, I encountered the same problem with the bioperl  
> parallel module.
> has any one noticed that?
> any ideas how to solve that problem?

My suggestion would be to double check the parameters in the output
files to ensure they are identical. If so, it would be worth putting
together a reproducible example and asking at NCBI.

Practically, my suggestion is to set up BLAST to run locally. This
ensures you have control over the BLAST version and search database
for reproducible results.

Hope this helps,
Brad