Believe it or not: how much can we rely on published data on potential drug targets?