Abstract 4938: A global landscape of transcriptional interactions and regulations in lung cancer

Proceedings: AACR 103rd Annual Meeting 2012‐‐ Mar 31‐Apr 4, 2012; Chicago, IL Lung cancer is the most common cancer in the world and is the leading cause of cancer-related deaths. Despite of decades’ intensive research, only a few driver genes have been identified and the exact regulatory mechanisms of lung cancer development and progression still remain elusive. The poor understanding of this deadly cancer prevents us from developing effective methods for early detection and personalized therapeutics. This study aims to reconstruct multiscale genome-wide transcriptional networks to elucidate the global patterns of transcriptional interactions and regulations for development and metastasis of lung cancer and to provide new insights into the disease through discovery of novel pathways and drivers. We compiled multiple large scale genomic studies of lung cancer. Gene coexpression networks were first reconstructed to identify gene modules (clusters) comprised of highly interacting genes. A module with unknown function along with the modules enriched for extracellular matrix organization, immune response, epidermis development and DNA replication/mitotic cell cycle are most predictive of survival. Meanwhile, causal Bayesian networks were also constructed to derive a more granular view of the relationships and directionality between genes. The coexpression and Bayesian networks were then integrated by projecting the co-expression modules onto the BNs to identify putative causal regulators (key drivers) within each module based on their interconnectivity and predicted impact on the rest of the module. About 200 driver genes including known oncogenes ATRX, BRCA1, BUB1B, CREB1, LCK, RECQL4 and SFPQ, were revealed. To further identify genomic alterations and key pathways account for metastasis of lung cancer, we applied a recently developed Wavelet based Analysis of Copy number alteration by Expression (WACE) algorithm to infer copy number variation sites and combined it with Bayesian networks to identify putative key drivers of the genomic alterations in metastasis. One hundred forty nine amplified regions and 91 deleted regions were identified. The amplified and deleted regions are most significantly enriched for gamma-aminobutyric acid signaling and MHCII-mediated immunity pathways, respectively. Twenty one key drivers including CHEK1, a key gene in DNA damage pathway associated with multiple cancers, and NME1, a known metastasis suppressor gene in breast cancer, were identified. The amplified regions are most enriched in the module with unknown function and they share three drivers (ACTR6, CEP68 and RPAP3). To our knowledge, this is the first effort to systematically reconstruct multiscale transcriptional networks and identify drivers for lung cancer. The global map of the gene-gene interactions and regulations as well as the inferred driver genes for both development and invasion of lung cancer will have a significant impact on lung cancer research and therapy. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 4938. doi:1538-7445.AM2012-4938