Post by nick777 on Jun 8, 2021 14:11:55 GMT
I have run several GCTA cojo-slct analyses on a number of traits and I'm interested in seeing how well the model selection in GCTA is replicated when using individual-level data.
I am replicating the GCTA results in R using the same selection procedure as described in Yang 2012: Conditional and joint multiple-SNP analysis of GWAS summary statistics identifies additional variants influencing complex traits.
My question surrounds the collinearity estimation between a variant to be selected and those already in the model. According to the paper above, this is done using the multiple regression R2.
I assume that the below R code would replicate the same collinearity estimation method as used in GCTA?:
r2 <- lm(new_variant ~ variants_in_model)$r.squared