Can the Gnomad PCA Be Used for Exomes

General
1

I have exome vcf data and would like to use the PCA from gnomad to determine the ancestries of my samples?

The following blog discussing PCA says:

If your dataset is missing large numbers of the variants with gnomAD PCA loadings, it will produce unreliable results.

Does that mean exome data cannot be used? Will also ask this question on the Hail forum in case it is something they would know.

Thank you.

2

Hi @beneopp,

Thank you for your patience! The PCA scores can be used for exomes when your exome sites are well covered by the PCA loadings themselves which cover 175,043 sites. You may want to do a quick check to ensure there is sufficient overlap between datasets. Also, you are already doing this by reaching out here, but please exercise caution when classifying using the gnomAD RF and consider the other caveats in the blog post.

Best,
Mike