search

metasnv-tool / metaSNV

MetaSNV, a metagenomic SNV calling pipeline.
http://metasnv.embl.de/
Other
19 stars 3 forks source link

metaSNV2 giving results for only 1 genome out of a 48 in a custom database #8

Closed hdore closed 2 years ago

hdore commented 2 years ago

Hello,

I'm having an issue with metaSNV2 that I can't understand. I'm running metaSNV2 on 90 samples mapped onto a custom reference database of 48 genomes (each genome has between 1 and 200 contigs, but most have less than 100 contigs).

metaSNV.py runs without errors, but it outputs SNVs for only 1 genome (and sometimes the first kb of another genome). Since this genome has a high coverage, it's alone in its split when I use 12 cpus (12 splits), but I don't get any result for other genomes that are also alone in their split. I tried changing the number of cpus (and even to use only 1 cpu) but it did not help.

The only messages I have in the log are 

Identified 90 samples
Found reference genomes and annotation file.
Loading Genomes...
Genomes loaded!
Weird... _Contig name_

With Contig name being the name of the first contig in the split. The only split for which I don't have the "Weird..." message is the one that worked.

I've tried to look at snpCaller script, but I can't read C so I don't know really understand how it works and where the issue could be.

The only difference I see between the genome that worked and all other genomes is that the one that works has no _ (underscore) in its contig name, while all others do. Do you think that could be the issue? Otherwise what else could cause that problem?

Other than that, I'm very happy with metaSNV2 results on the genome for which it worked!

Thank you for your help,

hdore

hdore commented 2 years ago

Hello,

I found the issue. All the genomes except the one that worked had a "Description" field in their fasta sequence header (that is, some information after a first space). After removing the description (i.e not allowing a white space in the sequence header), metaSNV2 worked as expected.

As my problem is solved, I will close the issue.

hdore