Table of Contents
Metanova Labs is developing a hit-to-lead optimization agent with Chutes and Onepot AI, extending NOVA, its drug-discovery platform on Bittensor’s Subnet 68, from finding promising molecules to refining them for further development.
The agent incorporates an algorithm developed by a winner of NOVA’s Blueprint competition. Researchers will be able to guide it through alternative optimization strategies, explore chemical modifications, and prioritize candidates using predicted binding affinity, selectivity, physicochemical properties, and specific interactions with a biological target.
Hit-to-lead optimization begins with a promising initial compound and works toward a more suitable starting point for drug development. Improving how strongly a molecule binds is only one objective. Researchers also need to consider whether it binds selectively, has suitable chemical and physical properties, and can be synthesized. The agent is being built to help navigate those competing requirements.
Onepot AI supplies proprietary combinatorial chemical databases spanning billions of possible compounds. These libraries describe molecules that can be made from combinations of chemical building blocks, connecting computational exploration to synthesis. Metanova says Onepot can deliver compounds within five to seven business days, building on the companies’ existing robotic-synthesis partnership.
Chutes provides the confidential computing infrastructure. Its team has integrated domain-specific models so inference, the process of running a model to produce predictions or responses, can take place inside verified trusted execution environments, or TEEs. These protected computing environments isolate sensitive workloads, with encrypted communication and controlled access to proprietary databases and project records.
That architecture addresses a practical requirement of pharmaceutical research. Compound structures, biological targets, emerging chemical series, and optimization strategies are valuable intellectual property. Metanova intends to use outside computing infrastructure while retaining organizational control over that research data.
The project connects algorithms produced through NOVA’s open competitions with proprietary research programs. Alongside its small-molecule work, Metanova has also begun nanobody production with Yalotein for experimental validation, giving NOVA a separate laboratory pathway for antibody-based candidates.
Metanova plans to integrate the new agent into its own hit-to-lead development program next, with the broader aim of supporting both its internal asset pipeline and external research partners. The optimization work precedes experimental activity testing and ADME validation, which examines how compounds are absorbed, distributed, metabolized, and excreted.
