Gravar-mail: On multi-marker tests for association in case-control studies