Abstract Unsupervised cluster analysis has been widely applied to omics data analysis for identifying molecular disease subtypes, which may present distinct disease prognosis and/or unique underlying disease mechanism. The findings can ultimately establish foundations for precision medicine. Existing disease subtyping methods in the literature are, however, purely unsupervised. The identified disease subtypes are often irrelevant to clinical outcome and cannot be translated into clinical practice. We hypothesize that an outcome-guided molecular disease subtyping framework with systematic integration of multi-omics data, clinical information and biological pathway knowledge will generate disease relevant and clinically actionable subtypes towards precision medicine. The developed methods are expected to be applicable for a wide range of complex diseases, where disease subtyping may be instrumental for finding novel therapeutic targets or improving treatment decisions. The specific aims are: (1) Develop outcome-guided clustering (OG-Clust) framework for disease subtyping using a single omics dataset. (2) Develop outcome-guided clustering (OG-Clust) framework for integrating multiple omics datasets. (3) Application and validation in breast cancer and pediatric asthma.