Plink p-value threshold from N on the X chromosome?

[kel...@umich.edu]

Hello,

With recent versions of Plink, the --hwe flag uses the method described in https://doi.org/10.1101/2024.02.07.24301951 to calculate a p-value threshold, which is partially based on the N for people genotyped at that SNP.

For the X chromosome, Plink uses the method described at https://doi.org/10.1038/hdy.2016.20 which includes information from hemizygous males, although I'm not sure if the resulting P-values will be similar to HWE p-values for the same sample size on an autosomal chromosome.

Do these two methods interact OK? Eg. is the N-based p-value threshold still appropriate for X chromosome SNPs when using an HWE p-value that includes data from hemizygous males? Or does Plink do something different to handle this?

Thanks!

Kristen

[Chris Chang]

You correctly suspect that, for a given p-value and sample size, the power of this test is a bit different on chrX than it is on the autosomes, and thus there should be a little bit to gain by setting this QC threshold differently for chrX.

The difference is less than it was with Plink 1.x, which ignored all the chrX male data.  Roughly speaking, you now have 25% less data to work with on chrX, instead of 50% less.

--
You received this message because you are subscribed to the Google Groups "plink2-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to plink2-users...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/plink2-users/f3527ee3-9957-41af-b74f-f2f8c0c4afa5n%40googlegroups.com.

[Kristen]

Thanks!!

So if the effective sample size is roughly 25% less, and I'd normally set k=0.001, then would k=0.00075 be appropriate for the X chromosome?

Kristen

[Chris Chang]

That's the setting I'd try, yes.

(Meanwhile, while we're on the subject of fine-tuning the HWE QC filter, have you looked at the 'keep-fewhet' modifier?)

To view this discussion visit https://groups.google.com/d/msgid/plink2-users/3b775bfe-be9d-4f22-aa47-1240e51dc61dn%40googlegroups.com.