Setting a threshold based on effective number of independent Snps in GWAS.

Question

[Abbas.M]

I want to set a threshold of GWAS based on effect number of independent snps in my genotype data. As i read in literature, most of the recently published papers have set threshold by this way. At first they calculated effective number of independent snps in their snp data. Then applied a basic bonferroni correction. Lets suppose, if effective number of independent markers are 20000 then the threshold would be like this, 0.05/20000, here 0.05 is the 5% error rate. Genetic type 1 error calculator can do it, but their website is unstable and i am unable to access it. Is there any alternative methods to estimate independent number of markers in a SNP data? Here is the link of the paper, in which they proposed this method, and other studies using their proposed method.

https://link.springer.com/article/10.1007/s00439-011-1118-2

https://science.sciencemag.org/content/sci/suppl/2017/01/25/355.6323.391.DC1/Tieman.SM.pdf#:~:text=The%20effective%20number%20of%20independent%20SNPs%20was%20calculated,parameters%3A%20-maxdistance%202000%20-minMAF%200.05%20-hwcutoff%200%20%2829%29.

https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1006889

Answer 1

[Nicolas Rosewick]

You could prune your dataset with plink : https://www.cog-genomics.org/plink/1.9/ld and then count how many SNP are in the resulting plink.prune.in file.

In Anderson et al. 2010 good GWAS practices paper, to compute a pruned dataset : https://www.nature.com/articles/nprot.2010.116

plink --file raw-GWA-data --exclude high-LD-regions.txt --range --indep-pairwise 0.15 50 5 0.2 --out raw-GWA-data

[Abbas.M]

thank you so much for answering my question. I did LD pruning in plink, but i was not sure about it, whether we can set a threshold of gwas on the basis of LD pruned snps. Once again thanks