Using EstWise in a large scale throughput manner

Estwise in a large scale manner is a more troubling issue than genewise. Generally the DNA databases are as large, but the algorithm is smaller and often people are equally interested in sensitivity and alignment quality. Therefore it makes more sense to use estwise directly as the database search. Estwise is still pretty slow, so here is a check list of things to do

• Run estwise on a clustered EST database, not raw reads
• Make sure you are using the 3:12 algorithm on estwise (-alg 312) for the database search. Try using the 3:12 quick (-alg 312Q).
• Use the pthread port if you have a large SGI/Sun/Dec multiprocessor
• For the final tuning, make sure you have switched on all the compiler options. egcs is a good thing to try. This will give you a 10-15at most

I am thinking about improvements to the estwise running time. I would very much like to collaborate with someone on estwise in terms of understanding its sensitivity and improving all aspects of the algorithm. Please get in contact with me.

The hardware solutions from Compugen, Paracel and Time Logic are all very good in this area, and worth investigating if you have money to spend.