Regularized regression using glmnet: No difference between groups?

Question

I am using a regularized regression to select several proteins that best discriminate health condition (binary: either disease or no disease). The purpose of using it is to