Why is Affymetrix computed sex different to that of plink?

Question

[HumeMarx]

Hi

I was wondering if someone can help me interpret the --check-sex results I am getting in plink.

When running this command (./plink --bfile file --check-sex 0.35 0.65 ycount out X) , I seem to get a lot of Problems being detected. I have done this a couple of times and ended up focusing on common snps only as I guess rarer snps can cause problems with F! Ran ycount at the same time (although I am only left with 5 SNPs on the Y chromosome) .

PLINK is detecting 58/162 problems in this relatively small cohort!!! The data came from an Affymetrix chip, and the axiom analysis suite is predicting the gender concordant with the pedigree data provided in all of these!

Here are some of the examples

(see image)

Any idea why this is happening?

Thanks

M

Answer 1

[HumeMarx]

In case anyone is having massive gender inconsistencies in Plink after exporting data from Axiom Analysis Suite: This was because there was an issue in converting the Affy files to PED/MAP format using the Axiom Analysis Suite. Regenerating the files using an upgraded (bug fixed) APT solves the issues.

[h.mon]

Which version(s) have the bug and which fixes it?

[Mari]

Hi I think I might be having the same problem, do you have more details of the APT version you used/conversion steps? Our suppliers initially did this for us but given the weird sex results I don't trust it.