[BioDarwin] BLASTN benchmarks

[David Adelson]

Corentin,

Feel free to download the query sequences and run them with WU-BLAST.  
We tried WU-BLAST on all of our machines but ran into problems getting 
it to run on the Xserves due to the 2GB memory limitation of that 
architecture.  We could get searches to run against Chromosome 21 but 
not against any longer sequences.   As  we would have had to break up 
longer database sequences we decided not to pursue the exercise, as the 
time spent fiddling around with sequence databases argued against any 
other gains in efficiency. NCBI (AG) BLAST is easy to set up and use 
and no special alterations of the databases is required.  For smaller 
databases or rather databases comprised of shorter sequences WU-BLAST 
might be a useful option, but it does not do what we need it to do 
without considerable fiddling around.

Cheers,

Dave


On Tuesday, September 2, 2003, at 10:20  AM, Corentin Cras-Méneur wrote:

> At 09:55 -0500 on 01/09/2003, you wrote:
>
>
>> I have run some comparisons of BLASTN searches between Apple Xserves 
>> (clustered and not clustered) vs large shared memory machines (sorry, 
>> no intel comparisons).  See http://blastcompare.tamu.edu for details. 
>>  Not surprisingly, the the Xserves were fastest at small word size.
>
>    It could have been interesting to perform the same tests with 
> WU-BLAST instead of AG-BLAST. The tests I ran a while ago showed that 
> WU-BLAST seemed to have a much better G4 optimization. I've heard that 
> it deals very well with multiple processors as well.
>
>
>
> 	Corentin
> _______________________________________________
> BioDarwin mailing list
> BioDarwin at bioinformatics.org
> https://bioinformatics.org/mailman/listinfo/biodarwin
>