no call in gvcf file of bacteria samples

[Negin]

I used GATK 4 for variant calling in bacteria samples. For some samples and some variants, there is no call. How can I know this no call is because of low-quality or gene absence?

[SNPsaurus]

I like breseq http://barricklab.org/twiki/bin/view...meResequencing because it shows the evidence for each variant call and highlights regions of no coverage as well.

[Negin]

Thanks @SNPsarus, but do you know also how can I solve it via GATK?

[SNPsaurus]

What is the output? A vcf file will typically give the genotype call and then report on read depth, allele depth, genotype probabilities and quality scores for the call, mapping, and sequences. Can you manually inspect the GATK output? You can also load the bam file into a genome viewer and inspect the region of interest.

[Negin]

@SNPsaurus, yes the output is a gvcf file (almost same as vcf file), and it has all the information that you mentioned. Here, is an example of one locus in one sample with no call.

eg. NC_000913 4473651 . C <non_ref> . . . GT:AD: DP:GQ:PL 0:2,0:2:0:0,0

and after combining with other samples and genotyping it with GASTK4, it will change to this:

GT:AD: DP:GQ:PL .:0,0:0:.:0,0

[SNPsaurus]

So this has 2 reads of the reference, but converts to a no call? Do other loci with only reads get called? There may be a firm threshold for needing >2 reads to support a call. GQ is 0, so that means alternative genotypes are equally likely. I'd say if other loci get called at 2 reads then you need to visualize the alignments to see what is going on.