Or use pymc 3 for the bayes equivalent. It's Better anyways. Yes one can also use rpy2. I think glmm is a gsoc project or an upcoming pr but not 100% sure.