Blast problem in metagenomics alignment

[rexxi]

Hi there, well I presented an annoying problem with blast. The problem is the next:

We are triying to aling a metagenomic sequence project with local blast, using software packages included in biolinux 6. but when it's running appears this error.

Error: basic_string::substr
xargs: blastx: acabó con estado 255; abortando
xxxx@xxxx-desktop[4-jul-2012] Error: basic_string::substr

We can't associate this problem to my computer who is equipped with a AMD phenom II 1090T, cause we are presenting the same trouble with 3 dell vostro who uses intel core i7 2600, all of them, with 8gb of RAM.

we tried to re-run this alignment again but the trouble continues.

We are running this with xargs, as is showed in this line. With 3000 sequences per file (1 file per core)

xxxx@xxxx-desktop[4-jul-2012] xargs -n 14 -P 3 -a arguments blastx

The error appears at 10 or 12 hours of started the alignment.

I'll be really thankful if somebody can give me an idea.

Sincerely, oscar

[Richard Finney]

code 255 is -1 ( http://tldp.org/LDP/abs/html/exitcodes.html )
blastx is abending (abortando).
"Error: basic_string::substr" is a C++ string overflow(?) error message.

Things you can do:

1) break down your input into more, smaller chunks. Isolate what input is causing the problem.

2) Download another executable for blast (blastx) from NCBI. This string overflow(?) could be a fixed problem.

3) Download the latest source code and re-compile and re-run using latest version.

4) Turn on "-g" (debug) as a gcc/g++ parameter in step 3 and use gdb to figure what's up. [ very advanced ] or use printfs to figure out what's up [ again, very advanced ]. Do-able but definitely for the bold and patient.

[jmeppley]

If you are creating your own blast-formatted databases, you may be able to fix this problem by using the "parse SeqID" option. With formatdb this is "-o T" with makeblastdb the option is "-parse_seqids".

[rexxi]

Thanks for your suggest, now I'm trying to finish this alignment, because it takes a lot of days, while I'll try the suggest of Richard Finney cutting where the alignment stops. For the next try I'll use the suggest of jmppley to ensure that my databases are not who gives me problems.

Thanks

PD: do somebody know why this troubles doesn't occur in our intel xeon servers? well, in these we have CentOS, not Ubuntu