We expect to see pathogenic variants, including dominant and biallelic recessives, as gnomAD is a population database where individuals with rare disease can be found, particularly in biobanks, and not a database of controls for which all diseases are excluded (except early lethal diseases). So the first step is to gather the prevalence of the disorder you are studying from the literature and determine if the allele frequency in gnomAD is above what you’d expect based on the population prevalence of the disorder. Only if the AF is above what you’d expect for prevalence, should you be concerned. That said, not all variants are real in gnomAD and if the AF is above what you’d expect for the prevalence, you should look carefully at quality metrics and the raw read data if available. If it appears high quality from that examination, then you might consider genotyping a set of samples that you have access to and performing orthogonal confirmation to understand if a given genotype is real. Unfortunately we do not have access to the samples from which gnomAD data derives, so can’t facilitate these studies ourselves.
1 Like