What Are Good Values For R^2 And D' For Linkage Disequilibrium

Question

[michealsmith]

I'm doing pairwise LD calculation between my GWAS SNP and the surrounding SNPs, and Ihave three SNPs here, which is around 10kb upstream of my GWAS SNP; and the following are the R^2 and D'

         R^2     D'
SNP1    0.39    0.89
SNP2    0.33    0.72
SNP3    0.39    0.89

I'm wondering does the data here show these three SNPs are in strong LD with GWAS SNP? Why D' looks quite good while R^2 is not that good?

thanks!

Answer 1

[coleman_jonathan]

Conventionally, R^2 of 0.8 is considered "tagging". The R^2 you have suggest imperfect LD.

The difference between the statistics is quite well explained here: PDF explaining LD

Basically, R^2 is a correlation dependent on D'. You will normally see higher levels of D than R^2.

[Aaronquinlan]

Your link is unlinked.

[coleman_jonathan]

Thank you. Fixed.

[coleman_jonathan]

Above link is dead: see this Biostars answer