Abstract
The role of genetic components in cancer development is an area of interest for cancer biologists in general. Intriguingly, some genes have both oncogenic and tumor-suppressor functions. In this study, we systematically identified these genes through database search and text mining. We find that most of them are transcription factors or kinases and exhibit dual biological functions, e.g., that they both positively and negatively regulate transcription in cells. Some cancer types such as leukemia are over-represented by them, whereas some common cancer types such as lung cancer are under-represented by them. Across 12 major cancer types, while their genomic mutation patterns are similar to that of oncogenes, their expression patterns are more similar to that of tumor-suppressor genes. Their expression profile in six human organs propose that they mainly function as tumor suppressor in normal tissue. Our network analyses further show they have higher network degrees than both oncogenes and tumor-suppressor genes and thus tend to be the hub genes in the protein-protein interaction network. Our mutation, expression spectrum, and network analyses might help explain why some cancer types are specifically associated with them. Finally, our results suggest that the functionally altering mutations in "double-agent" genes and oncogenes are the main driving force in cancer development, because non-silent mutations are biasedly distributed toward these two gene sets across all 12 major cancer types.