From: Spouge, John (NIH/NLM/NCBI) [E]
Sent: Monday, February 11, 2008 11:05 AM
To: NLM/NCBI List ncbi-seminar
Subject: CBB Seminar 11:00 am Tue Feb 12, B2 Library

CBB Seminar 11:00 am Tue Feb 12, B2 Library

 

"Real-time simulation of the statistical parameters for gapped BLAST" 

 

John Spouge

 

Biologists use the BLAST program more than once a second over the web to compare their query sequences to databases.
 If a query matches a database sequence of known function with a small p-value, the biological function of the query 
can be inferred. Presently, no on-line method can compute p-values to the accuracy the BLAST program requires, 
so sequence matches are restricted to certain pre-computed scoring systems. Over the past three years, my group has
 reduced (from about two days to less than one second) the simulation time required to estimate the BLAST statistical
 parameters, making on-line estimation possible.  The corresponding code is available in the NCBI CoreTools format.

 

In this talk, I will attempt to present (in readily comprehensible form) an overview of almost all the mathematical 
methods making this minor miracle possible.