UKBB_SNPTEST_Plink2

[Yuyang Luo]

Hi,

I am doing Hardy Weinberg for UKBB. I have tested for both Plink2 and SNPTEST. The chrX in UKBB is referred to non-PAR region, while chrXY is PAR region. I have two questions regarding this procedure:

1. What I am currently doing is applying HWE for chrX in Female only and applying HWE for chrXY in female and male together. Do you think it is correct in terms of chrX is non-PAR region and chrXY is PAR region?

2. Our results showed a big difference for Plink2 and SNPTEST regarding the HWE. Do you know the reason for the big difference? I would appreciate if any suggestions. Thanks!

Best,

Yuyang

[Christopher Chang]

From https://www.cog-genomics.org/plink/2.0/filter#hwe : "On chrX, p-values are now computed using the method described in Graffelman J, Weir BS (2016) Testing for Hardy-Weinberg equilibrium at biallelic genetic markers on the X chromosome."  This paper observed that on chrX, if the female genotype counts appeared to be in Hardy-Weinberg equilibrium, but male and female allele frequencies were different, the female genotype counts would be pushed out of Hardy-Weinberg equilibrium in the next generation.  As a practical matter, the joint genotype-count/allele-frequency test introduced by the paper has found obvious QC issues in data I've worked with, that were missed by the old female-genotypes-only method on chrX.

So, no, it is not correct to perform a female-only test on chrX.  Leave the males in.  Plink2 will account for them in an appropriate manner, as long as sex information is provided.

[Yuyang Luo]

Thank you so much. May I ask one more question which is how to use bfile or pfile to store dosage information from .bgen/.sample from UKBB? 

Best,

Yuyang

[Christopher Chang]

This automatically happens when you import bgen to pgen.  However, note that the dosage information is lost when you use --make-bed, since that format cannot store it.