Summary/Abstract Developing better therapies for complex diseases necessitates comprehensive understanding of intricate pharmacogenomic mechanisms. The explosion of multi-omic data and biomedical literature has enabled systematic explorations in pharmacogenomics; however, it is accompanied by substantial computational hurdles. Addressing this challenge, the PI’s laboratory has been pioneering state-of-the-art machine and deep learning models that comprehensively integrate diverse types of biomedical data to study disease biology, optimize treatment strategies, and ultimately enhance patient outcomes. We successfully applied our computational frameworks to diseases such as cancer, autoimmune diseases, hematopoietic disorders, and viral infections, yielding biologically meaningful insights. Over the forthcoming five years, the R35 award will augment the breadth and depth of our endeavors through three distinct yet synergistic themes: 1) predicting effects of therapies on diseased cells, 2) inferring pharmacogenomic interactions between genes and drugs, and 3) developing accessible computational resources. Specifically, Theme 1 will devise advanced deep learning models that integrate multi-omic information – ranging from genetics to transcriptomics and proteomics – to predict the molecular effects (e.g., inhibition of critical genes or pathogenic pathways) and phenotypic responses (suppression of cell activation, viability, etc.) induced by various genetic and chemical perturbations in disease models. By leveraging the emerging large language models, Theme 2 will dissect an extensive corpus of published literature to construct the landscape of pharmacogenomic gene–drug interactions. These interactions will illuminate the mechanisms of actions and molecular intricacies that govern treatment efficacy in the context of diseases. Theme 3 will create accessible computational resources that empower the utilization of cutting-edge computational methods and emerging genomic/pharmacogenomic profiling technologies. Completion of the proposed research will establish resources that facilitate cost-effective prioritization of therapeutic targets and agents for follow-up biological and clinical investigations, and evidence-based strategies for drug repositioning. Our research is innovative as it formulates a sophisticated computational framework that integrates deep learning machineries tailored to individual data modalities. The accessible tools will promote FAIR-ness (Findability, Accessibility, Interoperability, and Reusability) of relevant data. The framework established through this project is adaptable to computational methodologies and profiling technologies arising in the future, and broadly applicable across complex diseases. The PI is uniquely suited to lead the proposed research for his transdisciplinary experience in bioinformatics, engineering, and biomedicine, along with synergistic collaborations with wet-lab and clinical scientists in a vibrant tr...