Is it ok to replace missing WGS calls with reference notation "0/0"?

Question

[optimistsso4co3]

I called variants on 200 WGS samples, each got around 4 mil variants, however, most were unique and only 1 mil variants overlapped between most individuals.

I suppose it is normal behaviour that GATK won't output info about homozygote reference variants in genome, right? So the missing spaces should be simply filled with reference notation "0/0". Or shouldn't it?

Is there a correct way to fixt for low overlap? Otherwise it greatly complicates rare variant analysis.

Answer 1

[Pierre Lindenbaum]

So the missing spaces should be simply filled with reference notation "0/0".

no. If there is no read (e.g. an homozygous deletion), or if the quality is just too low then the call would be './.'

each got around 4 mil variants

there should be no "each" in gatk. Those samples should be called in GVCF mode and the combined with GenomicDBImport + GenotypeGVCFs

[optimistsso4co3]

1. Would GATK put 0/0 if reads contained only reference allele? Or is it the expected behaviour only in GVCF mode?

2. If i called them as a regular .vcf's is there a way to fix it and recall in GVCF mode apart from doing the whole thing anew?

[Pierre Lindenbaum]

1. no if the quality/depth is just too low.
2. no.