Calculating imputation accuracy Rsq for variants on chromosome X/23

[QinqinHuang]

Thanks a lot for this amazing tool!

I imputed my dataset in two batches using the TOPMed imputation server, because my sample size exceeds the maximum 25k. To combine the two batches, I am trying to calculate imputation accuracy estimates using all sample following the formula on your wiki.

I could replicate Rsq for autosomes using the HDS filed and ALT_AF, which is the imputed alternative allele probability. However, I seem to have issues with the X chromosome. For each variant, female samples have two HDS values from two haplotypes, while male samples have only one. How is Rsq calculated for X chr? Is it still average across all haplotypes? Then the number of haplotypes is 2N_female + 1N_male, not 2n in the formula.

Thanks a lot in advance for your help!

Best,
Qinqin