gnomAD AF = 0 vs gnomAD AF = NA (gnomad V2)

Question

[jon.klonowski]

Hello, I was looking at VEP annotated gnomad AF (gnomad V2, exomic, since I am analyzing exomes) and noticed that there was a class that was gnomAD AF = 0 and another set taht was gnomAD AF = NA.

Looking at these variants on gnomad, I noticed that the gnomAD AF = 0 variants had an entry in gnomAD that did indeed say the AF is zero. However for the gnomAD AF = NA variants, there is exome sequencing coverage for the area and there are even nearby variants but the page just said "variant not found"

https://gnomad.broadinstitute.org/variant/1-11769334-C-T?dataset=gnomad_r2_1 ^example of a AF = 0

https://gnomad.broadinstitute.org/variant/1-1293856-C-T?dataset=gnomad_r2_1 ^example of AF = NA

What does "variant not found" mean? whats the difference between this as AF =0.

Thanks so much ahead of time!

Answer 1

[jon.klonowski]

I got a credible explanation:

its probably because of the way GATK genotyping works. In that, areas without variation in the genotyping process will not be output by GATK, thus yielding an NA if one attempts to access those sites. But then what about 0 MAF then? Well, when one first does genotype calling there will be alot of calls, but a portion of those are false positives and are removed when the data is cquality controlled. These sites end up in your dataset a homo-reference; therefore, because there was an initial call, that position remains in the data as a homo-reference call.