How to calculate hemizygous counts

Browser
1

Hello!

I’m using Hail to build a pipeline that calculates allele frequency statistics by sex and ancestry. I’m aiming to replicate the gnomAD allele frequency table, which includes:

  • Allele Count (AC)

  • Allele Number (AN)

  • Number of Homozygotes (homozygote_count)

  • Number of Hemizygotes (n_hemizygous, when applicable)

  • Allele Frequency (AF)

All the statistics are outputed directly by Hail’s function hl.agg.call_stats but I want to ensure that I calculate correctly the hemizygous count, as I don’t see any direct way to do it.

My logic would be:

  • PAR regions: diploid → n_hom_var works as usual.

  • Non-PAR regions: haploid → we need to count any alternate allele in these samples as “hemizygous alternate.”

Is this how you get the hemizgous count shown in the gnomAD browser?

Thanks in advance!

Mireia

2

Hi Mireia,

That’s correct – in the browser pipeline, if a sample’s sex is XY and the variant is in a non-pseudoautosomal region of chromosome X or Y, then the any alternate allele in that sample is considered hemizygous alternate.

Also, note that in the data production pipeline upstream, we adjust each sample’s ploidy in non-pseudoautosomal regions using this function.

3

Hi Katherine,

Thank you very much for your reply! I’ll check the ploidy adjustion function and apply it to my code.