I-Corps: Translation Potential of a Hybrid Quantum-Classical Drug Discovery Platform

This I-Corps project is based on the development of a hybrid quantum-classical software platform for computer-aided drug discovery. A central bottleneck in bringing new medicines to market is accurately predicting how candidate molecules bind to the proteins that drive disease. This bottleneck slows the discovery of treatments for cancers, infectious diseases, and rare disorders, and contributes to the high price and slow pace of new therapies. This technology harnesses newly available, utility-scale quantum computers together with classical high-performance computing to model these interactions more accurately, while fitting into the familiar software tools scientists already use. More reliable predictions may reduce wasted experiments, shorten development timelines, and lower the cost of discovering new drugs, benefiting patients. This technology may be used by pharmaceutical developers, contract research organizations, diagnostic and precision-medicine companies, and quantum computing service providers, addressing a large and growing market and establishing a practical, high-value use for emerging quantum hardware. This I-Corps project utilizes experiential learning coupled with first-hand investigation of the industry ecosystem to assess the translation potential of a hybrid quantum-classical framework for structure-based drug discovery. This technology reformulates the prediction of protein binding pockets, the short, local regions where drugs act, as a constrained conformational-landscape problem rather than a whole-protein folding task. This approach captures cryptic, flexible, and mutation-altered sites that leading structure predictors often miss. Binding-pocket fragments are encoded as backbone bond-vector systems and sampled using prior-guided, hardware-efficient quantum circuits executed on utility-scale quantum processors. The measured results are refined into physically interpretable structural ensembles through energy-based filtering, subspace diagonalization, and clustering. Hardware-executed benchmarks show this method recovers native-like local geometry and produces docking scores correlating strongly with measured binding affinities, exceeding widely used classical and machine-learning baselines on difficult targets. By offering standard interfaces that work with common drug-discovery software, the platform delivers greater accuracy on hard targets while letting scientists keep their existing workflows and requiring no quantum-computing expertise. 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.