We are performing analysis on our EPIC Array data and are running into some issues on the analysis pipeline for the Epic version 2.

Specifically, we wanted to know the source from where the CpG MAF is derived from, and where we can find the corresponding MAF for other ancestries (for example, we want to filter out >0.5 MAF for EUR, EAS, and AFR ancestries)?

We are sourcing the v2 annotation here but could not find where the cpg_maf are sourced from:

https://www.bioconductor.org/packages/devel/data/annotation/html/IlluminaHumanMethylationEPICv2anno.20a1.hg38.html

Thanks all for any insight!

Those all come from dbSNP by way of UCSC. See here, particularly lines 114 and onward.

Here is no MAF for a CpG. The MAF is for the variants that are within the binding region for a given probe. The MAF reported is the lowest frequency, so it could be from any of the various populations. There are somewhere around 169k SNPs (If I look at dbSNP151 that is), and rather than downloading a massive file from UCSC and parsing it, you might be better off using the GenomicScores package