|
Post by arkofmind on Sept 2, 2020 9:46:07 GMT
Hi, I am wondering if I can run gsmr on a subset of snps with disease signal <1e-02. Would this cause a bias in the result? My disease gwas sample size is small n1325. So I am wondering if it's ok to limit the gsmr on a subset of disease signals, which would give better beta estimation. The risk factor is from public data and has enough 5e-08 significant snps.
If I didn't subset the snps, the gsmr pval is 0.096, with nsnp=1198. After subsetting, it's highly significant 9.82479e-41, with nsnp=28.
Thanks much, Yue
|
|