search

yukt / MetaMinimac2

11 stars 4 forks source link

Meta imputation with TOPMed and HRC (liftover) #5

Open lamanda0227 opened 1 year ago

lamanda0227 commented 1 year ago

Hi I'm trying to meta-impute across TOPMed and HRC, using output TOPMed imputation server and Michigan imputation server. Since HRC is in hg19, I used Picard to liftover the files to hg38. However, when trying to meta impute, I got this warning:

Checking phasing consistency across input files ... 
Phasing is not consistent at SNP xxxxxxxxxxxxxxxxxxx

I made sure that both TOPMed and HRC were using Eagle v2.4 for phasing on the server, and picard liftover should not change the phasing result whatsoever. I'm just wondering, in this case, would it be safe to turn on the --skipPhasingCheck option?

yukt commented 1 year ago

Though they use the same algorithm for phasing, the results can still be different. You may consider phasing in hg19, then liftover the phased vcf to hg38, and directly use it as the input for imputation against TOPMed, so as to make sure that the phasing status is consistent.

The option --skipPhasingCheck is intended for the occasion when you use the same phased vcf file for imputation, which does not apply to your case.

lamanda0227 commented 1 year ago

Thank you very much for the reply. I read through your documentation again and am considering using the empiricalDose.vcf.gz file way. Do I still need to turn on the phasing option while imputing if I use the liftovered empiricalDose.vcf.gz file?