Using GeneWise in a large scale throughput manner

If you are analysing genomic DNA in a large scale manner, you might wonder what is the best way to use genewise. Genewise is very CPU expensive compared to other programs. Part of this is because I have concentrated much more on correctness of the algorithm, not its speed (it is probably about 2 fold slower than it could be optimally), but mainly this is because the algorithm is complicated and DNA sequence is generally very large. I do not believe that optimising genewise in the code will solve people's CPU problems.

For these reasons, I do not advise the serious use of genewisedb as a single executable for comparing DNA sequence to either Pfam or protein databases. For these cases I suggest using the halfwise and blastwise scripts. See the section on Halfwise and Blastwise (5

• halfwise compares a DNA sequence to Pfam using a Blast speed up
• blastwise uses a Blastx result to provide the database search and provides sequence alignments ontop of that

Another option is to get in contact with Paracel, Compugen or TimeLogic, all of whom may be able to sell you specialised hardware. Paracel has successfully ported genewise to their hardware with only a few minor changes to the method.