Abstract
Abstract Background Cancer is known to occur due to mutations of oncogenes or tumor suppressor genes, which alter a series of downstream signal transduction pathways at molecular level. Therefore, inspecting interactive behaviors of proteins in cancer cells and comparing them with those in normal cells to obtain cancer-perturbed protein network can shed light on how a normal cell transforms into a cancer cell. Results Rough protein-protein interaction networks of apoptosis in cancer and normal cells are constructed according to human yeast-two-hybrid datasets and websites. Owing to high false positive rates in these large-scale interactome, the nonlinear stochastic model, maximum likelihood parameter estimation and Akaike Information Criteria (AIC) are employed to truncate fake protein-protein interactions of rough protein interaction networks. By comparing protein-protein interaction networks of apoptosis between HeLa cancer cells and normal cells, we can obtain cancer-perturbed networks and gain insight into the mechanism of apoptotic network in human cancer, which helps discovery of cancer drug targets. Furthermore, static and dynamic hubs in protein-protein interaction networks in caspase activation can be derived, too.