Pre-imputation checks using 1000G data (hg19) for a hg38 VCF

Question

[graeme.thorn]

I'm trying to use the pre-imputation checks here https://www.well.ox.ac.uk/~wrayner/tools/ to check a vcf (on the hg38 assembly) on the 1000G phase 3 v5 data, which is hg19, before imputing using the MIS. Obviously, very few of the variants in the vcf are matching locations in the hg19 1000G data.

My question is, is there a version of the pre-imputation checks that works on a hg38 dataset, or is it possible to use a lifted version of the hg19 data for these checks?

I've tried GATK LiftoverVcf to convert my hg38 data back to hg19, but as the vcf contains indels, I'm getting a lot of rejected variants because of the differences in assemblies.

Answer 1

[bk11]

Since your data is in hg38 build, you can perform pre-imputation checks following the section in Usage with the TOPMed reference panel in https://www.well.ox.ac.uk/~wrayner/tools/. You do not have to perform liftover for these checks.

[graeme.thorn]

That seems to have sorted it. I was following an older protocol which used hg19 vcfs & the hg19 reference 1000G panel, so I'm going to update it.