CHM13, high-coverage imputation

[gevro]

Hi,
Are there reference files available to perform imputation on CHM13-v1 aligned data?

Also, would QUILT work on high-cvg (~35x) whole genome sequencing, to help find the very small number of missed variant calls?

Thanks.

[rwdavies]

Hi,

No, sorry there are not reference files that I'm aware of.

QUILT can work on high coverage WGS, in the sense that it can run, but I don't think it would be helpful in your situation. Most reference data (e.g. HRC, or 1000 Genomes) uses the same tech (I'm assuming you've got short insert paired end Illumina or similar), so you likely won't gain SNPs in uncallable regions (as the reference panel would miss the same sites), and the imputation would contain errors at least because of the different history since divergence of any long matching reference haplotype versus your current sample (gene conversion and mutations that arose on one haplotype and not the other since they last coalesced).

If you have high coverage, other approaches like mapping to population graphs and calling variants on them might be more useful (they were a big thing a few years ago, not sure how easy to use / how commonly used)

Best
Robbie