Data Processing, Analysis and Modeling Unit

Project Summary/Abstract: Data Analysis Unit The over-arching goal of the Data Analysis Unit for the Washington University Human Tumor Atlas Research Center (WU-HTARC) is to provide bioinformatics tools and processing/analysis infrastructure for in-depth analyses of the data generated in the Characterization Unit. Most importantly, we will integrate data across

both the methodological (omics/imaging/phenotypic analyses) and the dimensional (1D/2D/3D/time) spectrums into coherent and accessible tumor atlases for each of the three cancer types: GBM, PDAC, and BRCA/TNBC. At the basic level, each atlas will consist of first cataloging a variety of numerically-computed metrics for cell

types, including fractions (1D), density, dispersion, and location measures for individual cell types and Euclidean measures of spatial interspersedness of different cell types, e.g. immune and tumor cells (2D and 3D), and how these metrics change with time. We will then correlate this information with both genomic

analyses, such as mutation signatures, clonality, and significantly mutated genes/regions/pathways, proteomic and metabolomics analyses, and image-derived data. The core of the atlas will be a MySQL relational database that not only stores all collected data, but links them along these different dimensions. Users will

interface with the atlas through a sophisticated viewer/query browser-based web portal that will support both traditional text-based queries, as well as spatial-based queries (shape, feature locations, etc.). The cohesion among the three atlases, in terms of the spectrum of data used and the approaches of their construction, will

allow users to generate new types of hypotheses not now possible and to perform pan-cancer analyses to reveal commonalities and differences in the three representative solid tumors, and to potentially extrapolate these findings to other tumors.