The primary bottleneck in drug development isn't creating therapies but identifying the right targets. Regeneron built its massive genetics database to find rare, protective genetic mutations in humans, effectively de-risking the target identification process and aiming to improve the industry's low success rate.
Human genetics doesn't just provide a drug target; it often specifies the therapeutic approach required. Discovering a protective loss-of-function mutation immediately tells researchers to develop an inhibitor (like an antibody or siRNA), accelerating the path from target discovery to molecule design.
While public AI models are powerful, they risk becoming commodities when trained on the same public data. Regeneron's strategy is to create a durable advantage by training AI models on its unique dataset of millions of genomes, proteomes, and linked health records to deeply understand human biology.
Instead of traditional methods, Regeneron sequences millions of people to find "superhumans"—those with rare genetic mutations that protect them from diseases. By studying these individuals, they identify high-confidence drug targets that mimic these natural protections, aiming for a higher probability of success in development.
Regeneron's focus on diverse populations is a core research strategy. Key discoveries, like the PCSK9 heart disease mutation, were only possible because they were significantly more common in African Americans. This proves that diverse genomic data unlocks unique and powerful therapeutic targets that would otherwise be missed.
To scale its database from millions to tens of millions, Regeneron is moving beyond bespoke global studies. The new model involves large-scale partnerships with health systems and consumer data sources, using privacy-preserving tokenization to securely link genetic data with vast electronic health records.
Regeneron pursues therapies for ultra-rare diseases, even without a clear standalone business model. The strategy is to treat these programs as the "tip of the iceberg," establishing a technology platform and biological understanding that can then be expanded to treat much more common diseases.
Regeneron views genomics as a "blueprint" for long-term risk. In contrast, proteomics acts as a real-time "sensor" of the body's current state. Their research showed proteomic data was surprisingly more predictive than genetics for the near-term onset of hundreds of diseases, including cancer and heart disease.