Model-based and model-free clustering: a case study of protein expression data for breast cancer