getSex warning in minfi - what is the distinction between the two sex prediction methods?

Question

[rkb965]

Hi all!

I am very new to this.

I have EPIC 850k data on ~1500 samples with relatively balanced sex.

When I run minfi::getSex I get the warning:

An inconsistency was encountered while determining sex.

What is the difference between the sex0 and sex1 objects within the function? Why is sex0 returned? What amount of discrepancy between sex0, sex1, and my known provided gender would reflect a data quality issue?

And I get slightly different results when running this after various processing points. I am removing some problematic probes and samples, and then using preprocessNoob (and next wateRmelon::BMIQ for normalization, but then I no longer have a rgset/mset). Where in the pipeline is it most appropriate to calculate getSex?

Thank you!

sex0 <- ifelse(dd < cutoff, "F", "M")
sex1 <- ifelse(k$cluster == which.min(k$centers), "F", "M")