Gravar-mail: Is protein classification necessary? Towards alternative approaches to function annotation