For each target SNP, GCTA uses the simple regression approach to search for SNPs that are in significant LD with the target SNP.
--ld ld.snplist Specify a list of SNPs.
--ld-wind 5000 Search for SNPs in LD with a target SNP within d Kb (e.g. 5000 Kb) region in either direction by simple regression test.
--ld-sig 0.05 Threshold p-value for regression test, e.g. 0.05.
Example gcta64 --bfile test --ld ld.snplist --ld-wind 5000 --ld-sig 0.05 --out test
Output files 1) test.rsq.ld, summary of LD structure with each row corresponding to each target SNP. The columns are target SNP length of LD block two flanking SNPs of the LD block total number of SNPs within the LD block mean r2 median r2 maximum r2 SNP in highest LD with the target SNP 2) test.r.ld, the correlations (r) between the target SNP and all the SNPs in the LD block. 3) test.snp.ld, the names of all the SNPs in the LD with the target SNP. Note: LD block is defined as a region where SNPs outside this region are not in significant LD with the target SNP. According to this definition, the length of LD block depends on user-specified window size and significance level.
Method for LD calculation: Yang et al. (2011) Genomic inflation factors under polygenic inheritance. Eur J Hum Genet. 19(7): 807-812. [Pubmed ID: 21407268]
GCTA software: Yang J, Lee SH, Goddard ME and Visscher PM. GCTA: a tool for Genome-wide Complex Trait Analysis. Am J Hum Genet. 2011 Jan 88(1): 76-82. [PubMed ID: 21167468]