How To Use Plink To Compute The Ld Between Two Sets Of Snps?

Question

[Fayue1015]

I have genotype data and I actually want to know the linkage relationship between two genes? So here is how I did, I map the snps which are reside to genes within +-10kb, then for two genes, each has an associated set of SNPs, how do I use plink to compute the LD between two sets of SNPs? Can I make an conclusion like this? If snp set A mapped to gene A has an average LD>0.8 with snp set B mapped to gene B, I would say gene A and gene B are linked? I find plink can only compute all pairs of LD for a given snp list. Thanks for any advice

Answer 1

[chrchang523]

This isn't directly supported by plink. Here are two ways to reduce the filtering workload, though:

1. Add --extract [SNP set {A+B}] --ld-snp-list [SNP set A]. It'll only be necessary to remove A-A pairs from the final report in that case.
2. Add --extract [SNP set {A+B}] and request matrix output. The values you care about will then all be in the lower-left (or upper-right) part of the matrix.

Answer 2

[Davy]

If you combine your two sets of SNPs into one list, and compute all pairwise LD, then simply remove all the LD values from pairs resulting from two SNPs in the same set.

You'd be best of trying something like this however with an LD mapping tool, like Haploview, and LocusZoom. How many SNPs in set A and set B are we talking about?

[Fayue1015]

set A and set B may contain less 100 snps, I will try Haploview and LocusZoom I do not want to use plink to do all pairwise than remove LD, too trouble some

[Fayue1015]

Hi, are you sure Haploview or LocusZoom has such functions? It seems do not have such function to me