LD Pruning a list of SNPs

[Nathan Lawlor]

Hello,

I have been looking through the plink 1.9 documentation and saw it was possible to calculate LD scores between a list of SNPS with the “--ld-snp-list” option and also perform LD-pruning of SNPs using the "--indep" , "--indep-pairwise" and other options. However, is it possible to perform LD pruning for a list of given SNPs? 

In the past, I performed LD pruning of SNPs using a VCF file containing genotype information for all chromosomes from the 1000 genomes phase 3 release (http://www.internationalgenome.org/category/release/). Note the VCF file contained several millions of SNPs and was a very large file (>15 GB). 

First I generated plink format files from the VCF:

plink --vcf all.chr.concat.vcf.gz --maf 0.05 --recode --out all.chr.genotypes --keep-autoconv

Then I used the independent pairwise option on .ped/.map files using the command:

plink --file all.chr.genotypes --indep-pariwise 1000 kb 5 0.2 --r2

Which would give me output files of all SNPs in LD that were removed (“prune.out” file) and all SNPS not in LD (“prune.in” file). I was wondering if it would be possible to obtain similar “prune.in” and “prune.out” output files while using the “--ld-snp-list” option. 

The type of analysis I was thinking of doing was something along the lines of:

plink --file all.chr.genotypes --indep-pariwise 1000 kb 5 0.2  --ld-snp-list Type_2_diabetes.SNPs.txt --r2 --out T2D.LD.SNPs

Where the “Type_2_diabetes.SNPs.txt” contains a list of diabetes GWAS SNP identifiers:

rs864745

rs12779790

rs7961581

rs7578597

…

However, when I run this analysis, it performs the LD pruning considering all SNPs in my .ped/.map files and does not restrict the pruning the SNPs in “Type_2_diabetes.SNPs.txt”. 

I would appreciate any help/feedback and please let me know if I can provide more information.

-Nathan

[Christopher Chang]

Separate the --indep-pairwise step into its own run (don't combine it with --r2), and replace "--ld-snp-list" with "--extract".

[Nathan Lawlor]

That worked great! Thank you for your quick response.