Towards the Identification of Disease Associated Protein Complexes

Motivation: Protein complexes play important roles in cellular functions. Many protein complexes have been identified by computational and purification methods so far. In which, cellular functions of protein complexes were well annotated; however, their roles in term of causing disease were not paid enough attention. Therefore, we here propose a method to identify associations between protein complexes and diseases.ResultsTo identify associations between diseases and protein complexes, we first constructed a protein complex network where two protein complexes are connected by shared genes. And then, we apply random walk with restart (RWR) algorithm on that network to rank candidate protein complexes based on their relative importance to known disease protein complexes. A leave- one-out cross-validation method was used to assess the performance of the method. As a result, the performance in term of area under the ROC curve (AUC) was archived at a relative high value (> 0.96). We tested our method with breast cancer and selected top 20 highly ranked protein complexes. Interestingly, ten of the selected protein complexes were evidenced to be associated with breast cancer.