[Bioperl-l] blastall and blosum62 problem

[Jason Stajich]

This is really a question for NCBI blast folks not us, but we'll still try
and be helpful.

Have you defined a .ncbirc according to the directions in the README.bls
this is typically where these problems come from.

-jason

On Tue, 10 Jun 2003, Mariano Latorre A wrote:

> I have a standalone blastall and a html version of blast on my server
> and i am having problems with my results:
>
> When I try the HTML version, I submit this pair of sequences
> ("sequences.in"):
>
>  >Test1
> AGCTTTTCATTCTGACTGCAACGGGCAATATGTCTCTGTGTGGATTAAAAAAAGAGTGTCTGATAGCAGC
> TTCTGAACTGGTTACCTGCCGTGAGTAAATTAAAATTTTATTGACTTAGGTCACTAAATACTTTAACCAA
> TATAGGCATAGCGCACAGACAGATAAAAATTACAGAGTACACAACATCCATGAAACGCATTAGCACCACC
> ATTACCACCACCATCACCATTACCACAGGTAACGGTGCGGGCTGACGCGTACAGGAAACACAGAAAAAAG
>  >Test2
> CCCGCACCTGACAGTGCGGGCTTTTTTTTTCGACCAAAGGTAACGAGGTAACAACCATGCGAGTGTTGAA
> GTTCGGCGGTACATCAGTGGCAAATGCAGAACGTTTTCTGCGTGTTGCCGATATTCTGGAAAGCAATGCC
> AGGCAGGGGCAGGTGGCCACCGTCCTCTCTGCCCCCGCCAAAATCACCAACCACCTGGTGGCGATGATTG
> AAAAAACCATTAGCGGCCAGGATGCTTTACCCAATATCAGCGATGCCGAACGTATTTTTGCCGAACTTTT
>
> I try a TBLASTX with BLOSUM62 and I get good results.
>
> But I need to use blastall with the script version...so I try with this
> command:
>
> cat sequences.in | blastall  -p tblastx -d /usr/local/blast/data/db1 -M
> BLOSUM62 | more
>
> and I get this errors and warnings:
>
> [blastall] WARNING:  [000.000]  Test1: Unable to open BLOSUM62
> [blastall] WARNING:  [000.000]  Test1: BlastScoreBlkMatFill returned
> non-zero status
> [blastall] WARNING:  [000.000]  Test1: SetUpBlastSearch failed.
> [blastall] WARNING:  [000.000]  Test2: Unable to open BLOSUM62
> [blastall] WARNING:  [000.000]  Test2: BlastScoreBlkMatFill returned
> non-zero status
> [blastall] WARNING:  [000.000]  Test2: SetUpBlastSearch failed.
>
> But remember this: when I try blastn I get good results again:
>
> cat sequences.in | blastall  -p blastn -d /usr/local/blast/data/db1 -M
> BLOSUM62 | more
>
> I need this working first, then I need to code some perl scripts.
>
> thanks,
> Mariano
>
>
>
>
> _______________________________________________
> Bioperl-l mailing list
> Bioperl-l at portal.open-bio.org
> http://portal.open-bio.org/mailman/listinfo/bioperl-l
>

--
Jason Stajich
Duke University
jason at cgt.mc.duke.edu