The “v4.0 AN issue” document posted on the gnomAD website says the allele frequency (AF) for impacted variants is artificially high. Per the document: “The reported allele frequency (AF) does not benefit from the total number of exomes of the full gnomAD v4 cohort, and therefore AFs for these ultra rare variants will be higher than the actual AF.” v4.0 AN issue - Google Docs
Question: Is the Exomes Grpmax Filtering AF (95% confidence) for impacted variants also artificially high?
Yes, in many cases the group max may be artificially high. The group max genetic ancestry group may also be incorrect as we are not accounting for all the samples present in v4.0 at the moment. The addition of the missing exome samples in v4.1 (coming soon!) could add a large number of samples, and thus ANs, to a genetic ancestry group which could change the current group max.