Re: [Biskit-general] database location
Brought to you by:
graik
|
From: Thomas E. <Tho...@po...> - 2009-12-10 22:25:25
|
Raik, >>> Interesting. I don't know what's going on here. My (meanwhile a >>> bit outdated) >>> installation of blast and blast databases doesn't give this error. >>> It could be >>> that the blastclust version you are using has changed it's command-line >>> parameters (There is a message like that in the log file). Please try using >>> blastclust stand-alone with a bunch of fasta sequences. For example: >>> >>> cd biskit/Biskit/testdata/Mod/project/sequences >>> blastclust -i nr.fasta >>> >>> this is just to test that the program itself is running. >>> >>> Then it could be that the new version of the nr database gives many more >>> sequences for the test querry and this may sometimes crash blastclust. >>> >>> So it would be best if you could reproduce the error outside the >>> Biskit test case... >> >> >> That's the output: >> >> $ blastclust -i nr.fasta >> Dec 10, 2009 4:56 PM Start clustering of 14 queries >> [blastclust] WARNING: Unable to open BLOSUM62 >> [blastclust] WARNING: BlastScoreBlkMatFill returned non-zero status >> [blastclust] WARNING: SetUpBlastSearch failed. >> [blastclust] ERROR: Failed to process query number 0 > > looks like you don't have a properly setup .ncbirc in your home or > it is not used. See the instructions on the biskit web site: You > need a file that contains something like this: > > [NCBI] > DATA=/usr/local/src/ncbi/data > > The folder is the one that contains the BLOSSUM and other matrix > files (not the databases!). > > Alternatively you can quickly try to set > setenv BLASTMAT /usr/local/src/ncbi/data > (or export...=) I had a BLASTMAT and env variable but set to designate the ncbi/data directory created by apt-get. Then I read the section explaining that repository-packages are outdated and installed the stand-alone version without changing it. Another silliness of mine. Anyway, that fixed WARNING: Unable to open BLOSUM62 but I still get *** buffer overflow detected *** with search_sequences.py Thomas |