Joint decomposition of SCORCH mulit-omic data to explore the impact of SUD and HIV on reward circuitry and neuroinflammation

Project Summary Researchers around the world have produced hundreds of multi-omic datasets relevant to the effects of substance use disorders (SUD) and Human Immunodeficiency Virus (HIV) in the brain, and the volume of high-quality data continues to grow exponentially. Combined analyses of these data are needed to fully harness their potential to discover biological mechanisms and therapeutic targets, yet several barriers impede such efforts: (1) The data are spread across many different repositories and have been processed in incompatible ways; (2) Standard statistical frameworks are neither scalable enough nor flexible enough for this purpose; and (3) Biologists and clinicians with limited bioinformatics skills are unable to directly interact with the data or analytical results. Here, we propose an ambitious effort to overcome these barriers, conduct a joint analysis of hundreds of multi-omic datasets related to SUD and HIV, and democratize access for the entire research community. We will begin by assembling a comprehensive resource of harmonized multi-omic datasets related to SUD and HIV, combining datasets produced by the SCORCH consortium with hundreds of additional datasets profiling the transcriptomes and epigenomes of single-cells, spatial positions, sorted neurons and glia, and bulk tissues from the brains of humans, non-human primates, and rodents. Next, we will perform joint analyses across this entire data compendium using newly developed joint decomposition and transfer learning techniques to identify molecular signatures of SUD and HIV. Our approach is uniquely suited to combine large numbers of heterogeneous datasets and to prioritize specific genes and biological mechanisms for experimental follow-up. We will create a biologist-friendly, web-based data visualization and analysis portal, leveraging the existing NeMO Analytics platform. We will extend NeMO Analytics to include hundreds of multi-omic datasets related to SUD and HIV, as well as by creating an application programming interface so that computational biologists can access the resource for analyses in R and Python. Collectively, our project will lead to a holistic understanding of the molecular effects of SUD and HIV in the brain and resources that are findable, accessible, interoperable, and reusable for the entire research community.