[BiO BB] Re: PSI-BLAST query filter

Iddo Friedberg idoerg at burnham.org
Thu Mar 10 12:37:24 EST 2005


The first round of PSI-BLAST is actually a regular BLASTP, but which 
generates a position specific score matrix. (NOT a consensus sequence). 
The second round is an alignment to the PSSM, generateing a new PSSM, 
and so on, ad nauseum or convergence (i.e. no new sequences).

You are right, neither is a true multiple alignment.


Kevin Berney wrote:

>a clarification perhaps, but doesn't psi-blast give a
>list of pairwise alignments to a -- "consensus"
>sequence.  or maybe a weight matrix? 
>whereas blast would just give you a list of pairwise
>alignments to the original query sequence.  
>these ought to produce very different results.  though
>neither is a true multiple alignment.
>Kevin Berney
>Quoting Iddo
>OK, I still do not understand why blastpgp insists on
>filtering your
>sequence... which version are you using?
>Not that:
>1) Using blastpgp You won't get a multiple alignment.
>You will get a
>multiple pairwise alignment to the query sequence.
>Using ClustalW you will
>get a true multiple alignment.
>2) There is still an inherent problem when aligning
>low complexity
>3) If those two factors do not matter, then why not
>use blastall to build
>your alignment? Why do oyu need the iterations
>blastpgp offers?
>Do You Yahoo!?
>Tired of spam?  Yahoo! Mail has the best spam protection around 
>Bioinformatics.Org general forum  -  BiO_Bulletin_Board at bioinformatics.org

Iddo Friedberg, Ph.D.
The Burnham Institute
10901 N. Torrey Pines Rd.
La Jolla, CA 92037 USA
Tel: +1 (858) 646 3100 x3516
The First Automated Protein Function Prediction SIG
Detroit, MI June 24, 2005

More information about the BBB mailing list