[Biojava-l] BLAST (ncbi-blast, wu-blast, web-blast)

[Dr. Christoph Gille]

In the last weeks, we had many discussions on BLAST which shows that
BLAST is of much interest. To my knowledge, there is no class in
Biojava  for invoking Blast searches yet.

Therefore I would like to discuss the new BLAST API with you.
It is a  Java wrapper for local NCBI and
local WU-blast and for Web BLAST at  http://www.ebi.ac.uk .

Please have a look at the API and tell me your opinion.

Have I missed something, are the method names OK ?

http://www.charite.de/bioinf/strap/Scripting.html#SequenceBlaster

http://www.charite.de/bioinf/strap/biojavaInAnger_SequenceBlaster.html

Please send your suggestions.

Here a short description:
Implementations of SequenceBlaster produce XML output which can be
parsed with org.biojava.bio.program.sax.BlastLikeSAXParser.  There is
also a simple non-Biojava DOM based parser which is currently used only to
make a human readable output.

The Wrapper provides a cache to avoid that one and the same BLAST is
run twice.  If this is, however intended, the BLAST result must be
removed from the cache before invoking compute().

The implementations of NCBI and WU-blast evaluate the shell variables
BLASTDB and WUBLASTDB, respectively which point to directories were
the databases are located. This works even for Java1.4 where the method
getenv() is corrupted. The Java wrapper can thus determine, what
databases are available.

The API is thread save. You can perform the computation outside the
event dispatching thread.

Christoph