MiNet

MiNet propose a gene and pathway-based deep neural network for multi-omics data integration, to predict cancer survival outcomes. In MiNet, gene-based multi-omics features are generated by considering main and interaction effects of multi-omics data in the multi-omics layer. The multiomics features produce canonical gene expression in the gene layer. The hierarchical representations of biological processes of multi-omics, genes, and pathways are captured in MiNet. MiNet showed the outstanding performance to predict cancer survival outcomes with GBM patients. More importantly, MiNet provides the capability to interpret a multi-layered biological system. A large number of biological literature supported our biological findings from MiNet.

This study was published in ISBRA 2019:
J. Hao, M. Masum, J.H. Oh, and M. Kang^†, "Gene- and Pathway-based Deep Neural Network for Multi-omics Data Integration to Predict Cancer Survival Outcomes", International Symposium on Bioinformatics Research and Applications (ISBRA), Barcelona, Spain, 3-6 June. 2019 - Regular paper (Acceptance rate: 22.6%)

Citation

@InProceedings{10.1007/978-3-030-20242-2_10,
               author = {Hao, Jie and Masum, Mohammad and Oh, Jung Hun and Kang, Mingon},
               editor = {Cai, Zhipeng and Skums, Pavel and Li, Min},
               year = {2019},
               title = {Gene-and Pathway-Based Deep Neural Network for Multi-omics Data Integration 
                        to Predict Cancer Survival Outcomes},
               booktitle = {Bioinformatics Research and Applications},
               publisher = {Springer International Publishing},
               pages = {113--124},
               isbn = {978-3-030-20242-2}