Single-sample versus case-control sampling scheme for Positive Unlabeled data: the story of two scenarios