ML/AI Tools to Advance Genomic Translational Research (MAGen)
MAGen is a National Institutes of Health (NIH)-supported research consortium advancing the development, validation, and translation of machine learning (ML) and artificial intelligence (AI) tools to accelerate genomic translational research. Our goal is to improve how we predict whether, when, and how disease will manifest in individuals who carry pathogenic (disease-associated) genetic variants—by learning from multimodal data, including both genomic and non-genomic data. The goal is to develop ethically focused and data-driven approaches to enhance the ability to detect and treat human diseases.
Program overview
MAGen is a multi-site consortium supported by the NIH—including the National Human Genome Research Institute (NHGRI), the National Institute on Aging (NIA), the NIH Office of Data Science Strategy (ODSS), and the Office of Behavioral and Social Sciences Research (OBSSR). MAGen brings together expertise across genomics, clinical research, data science, and AI to develop models that integrate genetic information with clinical, environmental, and lifestyle factors.
A defining feature of MAGen is cross-site validation: ML/AI tools developed at one site are independently tested and evaluated by other sites using distinct datasets. This approach strengthens robustness, reproducibility, and generalizability across populations and health systems.
Core objectives
- Develop and cross-validate ML/AI tools that improve prediction of disease manifestation for individuals with pathogenic genetic variants.
- Advance ethical, legal, and social implications (ELSI) research to support responsible development and use of ML/AI in genomic medicine.
What MAGen does
MAGen focuses on building ML/AI tools that can help answer real-world questions that often remain unresolved after genetic testing—such as:
- Will this person with a known pathogenic variant develop disease?
- When might symptoms begin?
- How severe might the condition be?
- How do non-genetic factors influence risk and outcomes?
By integrating multimodal data and validating approaches across multiple institutions, MAGen aims to move beyond variant identification toward more precise, individualized risk prediction and translational impact.
Why MAGen matters
Genetic testing can identify pathogenic variants, but it often cannot yet reliably predict what will actually happen to a person over time. Disease risk and outcomes are shaped by more than DNA—clinical history, environment, lifestyle, and other contextual factors also matter. MAGen is built around a central translational challenge:
If someone has a disease-associated genetic variant, what will actually happen to them?
MAGen addresses this by developing validated ML/AI tools that uncover patterns across genomic and non-genomic factors—so discoveries can more effectively translate into real-world clinical and public health benefits.
Consortium structure
MAGen is composed of three research “Development Sites” and a Coordinating Center that supports technical, scientific, and administrative coordination.
Development Sites include:
- Vanderbilt Health
- PRISM (Broad Institute, Cedars-Sinai, Harvard University, Brigham and Women’s Hospital)
- GEAR (Boston Children’s Hospital, Mount Sinai, Columbia University)
Coordinating Center:
The MAGen Coordinating Center provides technical, scientific, and administrative coordination for the consortium and supports the development and implementation of shared data models and dissemination strategies.
- Vanderbilt Health
MAGen is also advised by an External Scientific Panel appointed by the NIH to provide independent scientific guidance.
