2
$\begingroup$

I'm receiving a data set to start an analysis. My collaborators ran GATK to find variants from sequencing data, and GNOMAD calculated the minor allele frequencies - and then I'm receiving this MAF data to start the analysis.

When I was looking at what my collaborator called the minor allele frequencies returned by GNOMAD, some of the MAF were >50% (e.g. a MAF could be 70%).

I was looking at the GNOMAD documentation and it says this: https://gnomad.broadinstitute.org/help/faf but I'm not sure I understand what it's saying. Does it make sense that a MAF can be >50%? or are the allele frequencies returned by GNOMAD just the allele frequencies of the reference variant, and extra processing is requiring to find the MAF?

p.s. I can see this question. I actually don't have the data set yet, he was just showing me it and running through what the columns were yesterday, but the column looked like this:

GNOMAD.all MAF (i.e. all ethnicities MAF from GNOMAD)
SNP1    0.75
SNP2    0.002
SNP3    0.4
SNP4    0.1
$\endgroup$
1
  • $\begingroup$ Could it be that the numbers you saw were showing the frequency as a percentage rather than a [0,1] ratio? $\endgroup$
    – mgkrebbs
    Feb 4, 2022 at 19:13

1 Answer 1

2
$\begingroup$

Minor allele critically depends on what the reference population or genome is. And it also depends exactly how those data are sliced, it would be easier to say if we saw them.

Especially in large multipopulation consortia or aggregation efforts like gnomAD, this can be a pretty important consideration.

If your major allele is defined on something like 1000 genomes or, worse, a nation-specific population, then your "major" allele might quite often have frequency <50%.

These large data-collation efforts don't flip the major/minor allele around constantly, as that would require a significant informatics effort, and you would constantly have to track which major alleles line up with the reference genome and which don't. Probably they do update their major alleles periodically as more data gets sampled, genome builds get updated, and their curators get around to it, but it's unsurprising that at some frequency a specific population might have a minor allele at 80% or something.

It's a little more surprising when globally a minor allele is at high frequency, but it will still happen at some rate.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .