LD filtering

[Elielson Veloso]

Hello!!

I was wondering if it is possibel to run --r2 analysis for linkage disequilibrium and have PLINK to generate new binary files with SNPs with r2 passing a cuttoff threshold (e.g 0.2). I tried to run it as : 

plink --file MYFILE --r2 --make-bed  --out MYOUTPUT

My original .map file contains ~ 600 variants. After running --r2 command alone, my plink.ld file shows me all the variants with r2>0.2 and it has 196 lines ( not necessarily 196 variants, but surely less than 600 variants) I'd like to generate new binary files containing only those variants that passed the 0.2 cutoff for LD.

Would someone be able to help me?

Thanks in advance!

[Christopher Chang]

Taken literally, this request doesn't quite make sense, since r2 is a statistic calculated on *pairs* of SNPs, not individual SNPs.

However, plink has "LD pruning" commands which perform the high-level operation you're looking for; see --indep-pairwise in particular.  (Note that plink 2.0's implementation of --indep-pairwise is safer than plink 1.x's, because plink 2.0 ensures that there are no duplicate variant IDs that would make the followup --extract/--exclude operation do the wrong thing.)