Identification of gene functional modules shared by cancers based on biclustering

• Published in: Yíchuán Vol. 35; no. 3; p. 333
• Main Authors: Zhang, Fan, Lin, Ai-Hua, Lin, Mei-Hua, Ding, Yuan-Lin, Rao, Shao-Qi
• Format: Journal Article
• Language: Chinese
• Published: China 01.03.2013
• Subjects: Computational Biology; Databases, Nucleic Acid; Gene Expression Profiling; Gene Expression Regulation, Neoplastic; Gene Regulatory Networks; Genetic Pleiotropy; Humans; Multigene Family; Neoplasms - genetics
• ISSN: 0253-9772
• DOI: 10.3724/SP.J.1005.2013.00333

Pleiotropy is a common phenomenon in the genetics of cancers, which is rarely systematically evaluated. A novel idea for identifying shared gene functional modules using biclustering was proposed in this paper to explore the common molecular mechanisms among cancers and the relationships between different types of cancers. Gene expression datasets for 20 cancers were obtained. And genes differentially expressing in at least two types of cancers were selected using both moderated t-statistic and fold change to construct a 10417 × 20 matrix (gene-cancer matrix). 22 gene clusters shared by cancers were found by using the biclustering method. Further, Gene Ontology (GO)-based enrichment analysis identified 17 gene functional modules (Bonferroni corrected P < 0.05). The involved biological processes primarily included regulation of chromatids separation during mitosis, cell differentiation, immune and inflammatory response, and collagen fibril organization. These modules undertook molecular functions of ATP binding and microtubule motor activity, MHC class II receptor activity, endopeptidase inhibitor activity and so on. And their activity sites were mostly located in cytoskeleton, chromosome, MHC protein complex, intermediate filament, fibrillar collagen and so on. The network constructed based on these modules indicates that gastric cancer, ovarian adenocarcinoma, cervical cancer and mesothelioma were highly relevant to each other. However, the molecular mechanisms of two hematologic malignancies (acute myeloid leukemia and multiple myeloma) seem very different from other cancers. It can be seen that gene functional modules shared by cancers are associated with many biological mechanisms, and similarities among cancers are probably attributed to cellular origin and shared carcinogenic mechanisms. The proposed method for analysis of pleiotropy in this paper will help understand the common molecular mechanisms for complex human diseases.