Vexed causal inferences in nutritional epidemiology—call for genetic help