search

DRL / blobtools-light

Light version blobtools package - NO LONGER MAINTAINED! DO NOT USE!
8 stars 8 forks source link

BLAST complaining about wrong format #2

Open mw55309 opened 9 years ago

mw55309 commented 9 years ago

Hi

make-blobs.py ends with this:

[ERROR] : BLAST output does not seem to be in the right format ('6 qseqid staxids bitscore ... ')

Here is my blast command:

ncbi-blast-2.2.31+/bin/blastn -task megablast -query MARC_ecoli_032615/MARC_ecoli_032615.2D.fasta -db /mnt/ark1/data/nr/nt -outfmt '6 qseqid staxids bitscore' -max_target_seqs 25 -culling_limit 2 -num_threads 16 -evalue 1e-25 -out MARC_ecoli_032615/MARC_ecoli_032615.2D.nt.blast

What am I doing wrong?

Thanks Mick

DRL commented 9 years ago

Hi Mick, the command looks good to me ... although I haven't tested the new blast+ version yet ... could you sent me one of the lines of your output?

and also the command with which you ran "makeblobs.py" ?

cheers,

dom

mw55309 commented 9 years ago

Hi Dom

It’s an odd one – the command runs fine when the BLAST job is half way through, and the command finishes (despite BLAST still running). But then fails when I have full BLAST results.

Possible my version of nt and taxonomy are out of date – could that be the problem? What throws that error?

Cheers Mick

From: Dominik R Laetsch [mailto:notifications@github.com] Sent: 07 July 2015 16:54 To: DRL/blobtools-light Cc: WATSON Mick Subject: Re: [blobtools-light] BLAST complaining about wrong format (#2)

Hi Mick, the command looks good to me ... although I haven't tested the new blast+ version yet ... could you sent me one of the lines of your output?

and also the command with which you ran "makeblobs.py" ?

cheers,

dom

— Reply to this email directly or view it on GitHubhttps://github.com/DRL/blobtools-light/issues/2#issuecomment-119246843.

The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336.

DRL commented 9 years ago

Hi Mick, different versions of Taxdump/nt don't matter that much ... the program would only throw warnings in this case ...

It might have to do with the regex i use for blast-result parsing ..."^(\S+)\t(\S+)\t\s*(\S+)" ... or with the fact that second (taxids) and third (bitscore) columns might not be numbers ...

I now applied a hotfix that outputs the offending line when ERROR'ing during blast-parsing ...

Could you git clone the repo again and try it with the new version?

cheers,

dom