Figure 1
Workflow. 1:Input—the input data for KiMONo can be any mix of multiple omic data and prior knowledge. The prior represents general biological knowledge and is submitted via a list of already known associations between input features. 2:Prior based pre-feature selection—Based on the prior, KiMONo pre-selects omic features and generates a input matrix X for each gene. 3:Regression model—Each gene is modeled via a sparse group lasso using the genes expression as y and the previously selected matrix X as input. 4:Multi-Omic Network—all gene models are merged to generate a multi-level network containing features from all input sources as nodes and links for all non-negative regression coefficients between them.
