Towards an automated approach for protein identification in proteome projects