Ibd (Identity By Descent) In Control-Cases Study

Question

[eXpander]

Dear All,

My question is very simple I believe:

When performing IBD (with PLINK), should I do it separately for controls and cases, i.e. IBD in control-control and IBD in case-case pair. Or do I have to do it for case-control pairs as well?

[Matt Shirley]

Can you elaborate on what your goal is? Identity by descent for each of several SNPs in a case and control population to show that these SNPs are inherited in a family? IBD genome-wide to show that individuals are related to each other? IBD in windows to determine inherited blocks of SNPs?

[eXpander]

IBD genome-wide to find out duplicates and cryptic relatedness between individuals, used for Quality Control.

[Matt Shirley]

In this case I think you want to look at all pairwise comparisons between case-case, case-control, and control-control. One huge potential source for unexpected IBD is sample identity mix-up. Look at PMID 23185369 for an example of one strategy. PLINK is fairly dated, clunky, and insensitive. Consider looking at several new programs available for genome-wide IBD estimation.