Michigan Imputation Server caliberation

[wenjiezeng-cumc]

Dear authors, thank you so much for developing this great tool!

As you mentioned, "tsim was developed using output from the TOPMed Imputation Server v1.6.6 (Minimac4 for imputation, Eagle v2.4 for phasing, r2 for reference panel)", and recently you recalibrated the rsq to the TOPMed r3 reference panel.

I wonder whether I can use your tool on Imputed data output from the Michigan Imputation Server (Genome Asia Pilot as the reference panel)? Do I need to recalibrate the r2 filter or any other filters?

Thank you so much!
Wenjie

[Dongwon-Lee]

Hi Wenjie,
Sorry for the late response! In general, we believe that an R2 filter of 0.99 can be used without calibration. However, if there are not enough SNPs after merging or you want to lower the r2 filter, you should recalibrate it for the reference panel. For the other filters, especially ER2, we have found this unnecessary with the stringent r2 filter. See the discussion on #1