Collaborative Research: SCH: Protecting Privacy and Promoting Fairness in Advanced Genomic Research using Federated Learning

About This Grant

Recent advances in sequencing technologies have enabled collection of large-scale, multi-cohort genomics and genetics datasets. Among these, spatially resolved transcriptomics (SRT) offers the ability to measure gene expression across tissue sections while preserving spatial context, providing critical insights into cellular organization, and disease mechanisms. To fully realize the scientific potential of these datasets, integration of multi-cohort genomic and genetic data from different institutions is essential. Individual studies typically lack the breadth of biological and technical variation required for comprehensive analysis or robust model development. By jointly analyzing data from multiple sources, researchers can detect reproducible molecular patterns, strengthen the reliability of computational methods, and uncover signals that may only emerge in larger combined cohorts. Such Integrative studies also promote reproducibility through cross-validation across datasets and enable researchers to uncover new insights by reanalyzing existing data, thereby maximizing its value and reducing redundant collection efforts. As such, integrative analysis has become a cornerstone of modern biomedical research. However, current approaches to data integration often rely on centralized data sharing, which poses significant privacy and regulatory challenges. Both genomic and genetic datasets, such as those used in SRT and polygenic risk score (PRS) modeling, can contain sensitive information related to individual traits, health conditions, and ancestry. Sharing such data across different institutions raises serious concerns about confidentiality and compliance with data governance policies. Moreover, differences in infrastructure and access further limit the feasibility of centralized analysis. These challenges hinder the scale, consistency, and accessibility of collaborative studies, particularly in applications such as multi-omics data integration, and PRS prediction, where large and heterogeneous datasets are essential. Addressing these limitations requires new computational frameworks that enable collaborative analysis without exposing sensitive data. This project introduces FLAG (Federated Learning for Advanced Genomics), a federated learning framework for secure, scalable analysis of multi-institutional genomic and genetic datasets. The research includes three aims. First, the team will develop federated spatial representation learning methods that preserve fine-scale tissue structure and extract low-dimensional molecular features across institutions without data sharing to protect privacy. Second, the project develops federated Bayesian models to improve the accuracy and generalizability of PRS predictions from genetic data across heterogeneous cohorts. These methods incorporate hierarchical priors and uncertainty quantification to optimize model robustness across populations. Third, the project will release a user-friendly software platform that enables decentralized analysis workflows, allowing institutions with limited computational resources to participate in federated modeling without requiring centralized infrastructure. The proposed methods are grounded in rigorous statistical principles and tailored to the privacy, scalability, and structural demands of high-dimensional biomedical data. By enabling secure cross-institutional analysis without compromising confidentiality or requiring data centralization, FLAG offers a robust foundation for collaborative research in genomics and precision medicine. The framework is also applicable to other biomedical domains, including electronic health records and pathology imaging. Ultimately, this project will provide the research community with practical tools for privacy-aware genomic and genetic discovery, advancing reproducible science and enabling broader collaboration in data-driven biomedical innovation. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.