[Biojava-dev] Fwd: bioperl like blastparser

Michael Gang michaelgang at gmail.com
Mon Jan 7 10:16:22 UTC 2008


Hi Andreas,

You are correct.
The junit.jar library was missing in my ant_home.
Eclipse wrote that it was running the tests, but did not run any.
Now I corrected it and see that tests are failing.
I ran the program BlastEcho.java manually on the blast test files and
on the ncbi blast.
Judging after manually curation it worked good but at wu_blast id did
not parse the query length.
The reason is that in wu_blast the query length line has just 8 spaces
at the beginning instead of 9.
So I corrected the line which identifies the querylength at
org.biojava.bio.program.sax.BlastSaxParser
line 68 to: if (poLine.matches("^\\s+\\(\\d+\\sletters\\)\\s*$")) {

Now it works also on wu_blast.
It would be now a good idea to update the blast tests regarding the
number of arguments and see if the fail still.

Thanks in advance,
Michael



On Jan 6, 2008 2:41 PM, Andreas Prlic <ap3 at sanger.ac.uk> wrote:
> Hi Michael,
>
> I just had a look at your patch for the query length.
> Several of the unit tests are now failing at
> org.biojava.bio.program.ssbind.SSBindCase.testResultGetAnnotation(SSBindCase.java:143)
>
> The problem is that most blast related unit tests extend the SSBindCase,
> which expects a fixed number of attributes. With the new patch some of the
> blast-flavors have the additional queryLength attribute.
>
> Could you have a look at the behaviour of the parser for some of the files
> where the tests now fail? If you think the new behaviour of the
> parser is correct, we can simply update the tests to accept the different
> number of attributes.
>
> Thanks,
> Andreas
>
>
> --------------------------------------------------
>
> Andreas Prlic      Wellcome Trust Sanger Institute
>                     Hinxton, Cambridge CB10 1SA, UK
>
>
>
>
> --
>  The Wellcome Trust Sanger Institute is operated by Genome Research
>  Limited, a charity registered in England with number 1021457 and a
>  company registered in England with number 2742969, whose registered
>  office is 215 Euston Road, London, NW1 2BE.
>



More information about the biojava-dev mailing list