Extracting multi-function features for cancer genes

The causation of cancer often involves the joint deregulation of multiple biological processes. Thus, it is interesting to extract multi-function features of cancer genes and study their functional coordination involved in tumorigenesis. Here, based on Gene Ontology, we proposed a heuristics strategy to extract multi-function features which are significantly overrepresented with cancer genes. We showed that the extracted feature combinations can provide hints for studying the functional interplay roles of multi-function genes involved in tumorigenesis. Using protein-protein interaction data, we found that the pivot genes of the obtained functional feature combinations are likely to be cancer genes, and that their alternations may play important roles in simultaneously deregulating multiple functions and thus inducing cancer.