change: fixed make_s; combined -> __combined__

original experiment and experiment_id must be dropped when making s
so as to have a the same OD_mean for all experiment_ids.