Relation Therapeutics developed MORGAN, a foundation model trained on perturbation datasets to identify causal mechanisms rather than statistical associations in drug discovery. This shift from correlational to mechanistic understanding addresses a critical gap in how biomarkers are interpreted, with implications for targeting interventions across age-related diseases including immunology, metabolic dysfunction, and bone disorders.
Key Points
- MORGAN trained on perturbation data, not observational patterns alone
- Distinguishes causal mechanisms from statistical associations in biology
- Lab-in-the-Loop platform couples predictions with experimental validation
Longevity Analysis
Current aging biomarkers excel at prediction but remain silent on intervention—they describe correlations without revealing what perturbations would change outcomes. Relation's mechanistic approach addresses this directly by training models to understand cellular responses to deliberate biological stress and pharmacological challenge. As age-related diseases become prevalent with time, a framework that decodes cause-and-effect relationships in cellular behavior becomes foundational to identifying which therapeutic targets actually matter. The continuous cycle of computational prediction informing experiments, then experimental results refining the model, accelerates the transition from passive observation to active intervention.
Original published by Longevity.Technology, by Eleanor Garth.