? Data Analysis Unit We propose to create a Data Analysis Unit in service of the Omics and Multidimensional Spatial (OMS) Atlas. The OMS Atlas will enable discovery of mechanisms of resistance that arise in individual patients with metastatic breast and prostate cancer during treatment with current generation of targeted therapeutic combinations and immune checkpoint inhibitors. Treatment will these therapies in metastatic cancer is rarely effective for an extended period of time, and understanding the mechanisms by which these cancers become resistant to therapy is the primary goal of the OMS Atlas. The Data Analysis Unit will support this goal by developing and deploying data management, processing, analysis, and visualization methods and software to create the Atlas. The OMS Atlas will collect two biopsies, one before treatment and one during treatment for 3 different cohorts of cancer patients. The final product of the Data Analysis Unit will be a complete tumor atlas accessible via an interactive portal that enables use-case biologists in the OMS Atlas, the HTAN and the larger research community to develop hypotheses about tumor resistance mechanisms through quantified, longitudinal, and spatially-resolved comparisons of pre- and on/post-treatment biopsies from individual patients. Using primary data generated from omics and imaging assays (Tier 1 data), the Data Analysis Unit will generate three additional tiers of data: (Tier 2) single gene/cell measurements obtained by processing data from a single data platform; (Tier 3) tumor maps generated by combining single-cell and spatially-resolved omics and imaging data as well as quantification of systems-level functions such as biological pathway activity and the cells comprising the tumor and its surrounding tissue using integrative analyses of multiple data platforms; (Tier 4) a tumor atlas that can be used to compare pre- and on/post-treatment biopsies and identify features potentially correlated with resistance to treatment. Data tiers will be generated using a robust software pipeline consisting of a data management system, image management software, a workflow execution system, and visualization tools. Standardized and reproducible workflows that run on this platform will be implemented to generate all tiers of data. Statistical and machine learning approaches will be used to create tumor maps by connecting mirror image sections and cell populations across different assays. The OMS Atlas portal will provide a single interface with access to 10 different visualizations of tumor maps. Tumor maps can be visualized and compared longitudinally within a single patient or laterally across patients. Many visualizations can be displayed simultaneously using a dashboard approach where visualizations can be progressively added as desired, making it possible to view many different types of data about tumor maps simultaneously. Specialized animation approaches and 3D techniques will be used in visualizations to effectively display multidimensional, spatially resolved tumor map data.
