AMARETTO-Hub: a Knowledge Graph-based software platform that leverages the *AMARETTO software toolbox for multimodal and multisc

About this Class

We present the AMARETTO-Hub as a Knowledge Graph-based software platform that leverages Neo4j and Shiny to embed and interactively interrogate results generated by the *AMARETTO software toolbox that offers modular and complementary solutions to multimodal and multiscale network-based fusion of multi-omics, clinical, imaging, and perturbation data across studies of patients, etiologies and model systems of cancer and COVID-19, towards better diagnostic, prognostic and therapeutic decision-making in complex disease. For several Use Cases of cancer and COVID-19, we provide the biomedical research community with R Jupyter Notebook workflows that run the Bioconductor and GitHub repositories on Google Colaboratory, and GenePattern Notebooks that run the GenePattern modules in the Amazon Cloud, and that generate HTML reports comprising queryable tables with heatmap and graph visualizations in an automated manner, and additionally provide users with Neo4j-embedded Shiny interactive representation and querying tools that redirect users to *AMARETTO-generated HTML reports. Specifically, our software toolbox comprises of the following algorithms: (1) The AMARETTO algorithm learns networks of regulatory circuits - circuits of drivers and their target genes - from functional genomics or multi-omics data and associates these circuits to clinical, molecular and imaging-derived phenotypes within each biological system (e.g., model systems or patients). (2) The Community-AMARETTO algorithm learns subnetworks of regulatory circuits that are shared or distinct across networks derived from multiple biological systems (e.g., model systems and patients, cohorts and individuals, diseases and etiologies, in vitro and in vivo systems). (3) The Imaging-AMARETTO algorithm maps radiography and histopathology imaging data onto the patient-derived multi-omics networks for imaging diagnostics and prognostics to identify clinically relevant imaging biomarkers and decipher their underlying molecular mechanisms. (4) The Perturbation-AMARETTO algorithm maps genetic and chemical perturbations in model systems onto patient-derived networks for driver and drug discovery, respectively, and prioritizes lead drivers, targets and drugs for follow-up with experimental validation, towards better therapeutics. (5) The AMARETTO-Hub platform for Knowledge Graph-based embedding of knowledge learned via multimodal and multiscale network-based data fusion in previous steps. In these complex graphs, nodes and edges represent the diverse range of biomedical entities and the relationships between them, respectively. Graph-based embedding enables querying these complex graph-structured representations in a more sophisticated, efficient and user-friendly manner than can otherwise be accomplished by table representations alone. Resources are available from : github.com/broadinstitute/BioC2020Workshop_AMARETTO-Huband portals.broadinstitute.org/pochetlab/amaretto.html(to be updated). The ITCR Program is a trans-NCI program supporting investigator-initiated, research-driven informatics technology development spanning all aspects of cancer research. The ITCR Program funds tools that support the analysis of -omics, imaging, and clinical data, as well as network biology and data standards. All of the tools are free for use by academic and non-profit researchers. Access to tools, code repositories, and introductory videos are available on the website itcr.cancergov/