Closed Shicheng-Guo closed 2 years ago
The REGENIE method includes a pre-processing step which essentially identifies and remove covariates which are perfectly correlated (or with the intercept). So for a covariate like breast cancer where Sex is the same across individuals, Sex would be removed as it would be perfectly correlated with the intercept.
Hi Joelle,
In large scale pheWAS analysis, the covariates selection is a challenge. For example, for majority traits, gender should be adjusted, but for gender specific traits, like breast cancer, it is not a good idea to take gender as a covariate. I am wondering is there any smart way to achieve the automatic selection for such specific case in a large-scale phewas analysis?
Thanks
Shicheng