Argument: Doing pharmacogenetics is futile since the frequency of clinically relevant genomic variations does not differ between the patient population and the normal population.

Reply: Since the frequency of most of these genomic variations is high, you would need a large cohort to verify that no over-representation exists in your patient samples. In addition, a negative answer (no variation detected) is not the same as a useless answer, as it can be used to look for other reasons for the condition presented.