How is imputation r2 calculated in minimac3?

Question

[cityofwisdom]

After genotypes are imputed using Minimac3, an r-squared threshold is generally applied to weed out poorly imputed genotypes.

Can anyone please explain how this r2 is calculated?

The minimac3 wiki pages define r2 as:

an "estimated value of the squared correlation between imputed genotypes and true, unobserved genotypes. Since true genotypes are not available, this calculation is based on the idea that poorly imputed genotype counts will shrink towards their expectations based on population allele frequencies alone; "

Is the r2 based on the population tested or the reference panel (HRC/1000G/CAAPA)?

Answer 1

[chrchang523]

Minimac3's r-squared metric is computed off your dataset (not the reference panel), and is defined as [observed dosage variance] / [expected dosage variance, given observed allele frequency and assuming Hardy-Weinberg equilibrium].