LD Pruning with Plink-1.9

[sohail.l...@gmail.com]

Dear All,

I have Binary version of Plink BED file with variants from all chromosomes and of different individuals from multiple populations around the globe. I want to Perform LD pruning using Plink-1.9, with following parameters:

./plink --indep-pairwise 1000 50 0.2

//here's 1000 is windows size, 50 is step size and 0.2 is r^2 threshold. 

My questions is:

1. Should i have to split my file into different chromosomes, like chr1, chr2 and later perform LD pruning or is that Plink can handle variants data from different CHRs (i.e. autosomes as well as sex chromosomes) and variants from multiple population individuals together??

Thank you very much for your kind reply in advance!

[Christopher Chang]

Plink will automatically handle the chromosomes separately; there is no need to split the file.

[sohail.l...@gmail.com]

Dear christopher,

Thank you for your reply! can we run plink on multiple threads (i.e. parallel processing/cluster computing)?

[Christopher Chang]

Plink 1.9's LD pruning is not automatically multithreaded; if you need more speed, then chromosome splitting and then handling each chromosome on different processes/machines will work.  (If you're using different processes on the same machine, with e.g. GNU parallel, make sure to use the --memory flag.)

(Plink 2.0 includes an automatically multithreaded LD pruner.)

[sohail.l...@gmail.com]

Thanks Chris for useful info.

I used following commands:

../plink -bfile inputfile-name --indep-pairwise 1000 50 0.2 --out outputfile-name      /// this command generated only "pruned.in" and "pruned.out" sites 

./plink -bfile inputfile-name --indep-pairwise 1000 50 0.2 --out outputfile-name      /// this command generated bed file along with "pruned.in" and "pruned.out" sites. but the bed file is the same.

Purpose: i wanna generate PLINK BED file after excluding LD pruned sites, but in the second command it does not work. Could you please guide what is the correct way to do so?

another ou-of-context question: if i wanna generate BED file with genotype missing rate is less than 3%, how to use --geno option?? is it "--geno 0.3" or "--geno 0.03"?? I am confused a bit!

Thanks!

--

[Christopher Chang]

1. To generate a new pruned fileset, you need to run two plink commands in order:

plink --bfile inputfile-name --indep-pairwise 1000 50 0.2 --out outputfile-name

plink --bfile inputfile-name --extract outputfile-name.prune.in --make-bed --out outputfile-name

2. 3% is equal to 0.03, so you'd use --geno 0.03 in your --make-bed command.