LD analysis in PLINK based on reference and a SNP list

Question

I'm new to PLINK and genetics, and getting confused with two PLINK commands for LD analysis:

1. plink --bfile hapmap --r2 --ld-window-r2 --ld-snp-list --ld-window

2. plink --noweb --bfile --clump hapmap --clump inputfile --clump-field P --clump-p1 --clump-p2 --clump-r2 --clump-kb

It seems to me that for #1, the input files are a reference data and a SNP list file, while for #2, it is a reference data and a .assoc file.

In the output for #2, p-values are included with other results (according to PLINK manual) which is not the case with the output from #1. If I'm interested in finding proxy SNPs for the SNPS on my list in hapmap data, will it be sufficient for me to use the output from #1?

After invoking command #1 using PLINK 1.9, its output gave me 6 columns: "Chromosome number", "chromosomal position", "SNP name" for query/target SNP and 1 column of "R2". I cannot carry out command #1 because I do not have the .assoc file and only have a list of SNPs.

My main question is - what is the difference between these two commands and which of the two should give me the best result for finding proxy SNPs for my list of SNPs.

Answer 1

It looks like the first command is calculating pairwise linkage disequilibrium r2 between your SNPs.

The second command is clumping your dataset based on a p-value over a file of association test results, which I guess is a pre-cursor to computing polygenic risk scores.

Clumping is the process of thinning your SNP list to only keep uncorrelated SNPs so that each representative region has only one marker. In this case you are asking PLINK to keep the marker with the highest p-value.

I'm not sure what your definition of proxy SNPs is but if you are looking for tag SNPs to a SNP of your interest, you should go with command 1 and choose a proper LD r2 threshold.