search

niemasd / ViReflow

An elastically-scaling automated AWS pipeline for viral consensus sequence generation
GNU General Public License v3.0
10 stars 2 forks source link

Bug when there's ridiculously low coverage #12

Open niemasd opened 3 years ago

niemasd commented 3 years ago

This file was a dummy example run that only allowed 10 successfully-mapped reads (so pretty much all positions are coverage 0, with some coverage 1), and the test.low_depth.tsv file is fine (it says all positions are low depth), but the consensus sequence has nucleotides at pretty much all positions, whereas it should be N everywhere:

test.tar.gz test.rf.txt

Probably some (extremely unrealistic) corner case in the conversion from VCF to consensus sequence

niemasd commented 3 years ago

The bug seems to occur when the VCF file has no variants. If I put even one dummy variant in a low-depth region (i.e., in a region included in *.low_depth.tsv), bcftools masks the consensus sequence properly