Improving causality in microbiome research: can human genetic epidemiology help?