Fig. 1

Schematic workflow of MOSD. When integrating multi-omics datasets, MOSD first create the affinity using each data type where rows are patients and columns are gene features. MOSD then assign weights for the affinities and perform linear combination in integration step, followed by implementing self-diffusion to enhance the similarity of integrated network. Spectral clustering is used to obtain the patients labels with the optimal clustering number estimated by separation cost method. The identified subtypes are evaluated by survival analysis and biological analysis