Scalable Quantum-Enhanced Generative Models for Drug Discovery: Extending QCBM with NQS and CUDA Quantum for Multi-target Inhibitor Design

This project proposes a scalable quantum-enhanced generative model for drug discovery, targeting the design of KRAS inhibitors through the integration of Neural Quantum States (NQS), Quantum Circuit Born Machines (QCBMs), and CUDA Quantum acceleration. By replacing traditional quantum circuits with adaptive NQS representations and extending QCBM capacity to 32 qubits, the framework enables more expressive and efficient sampling of molecular space. Reinforcement-inspired optimization and virtual screening tools are incorporated to refine molecule generation toward biologically relevant candidates. The platform aims to overcome current limitations in scalability and generalizability of quantum models, offering a fast, adaptive, and physically grounded approach to multi-target drug design.

Faculty Supervisor:

Alán Aspuru-Guzik

Student:

Partner:

National Yang Ming Chiao Tung University

Discipline:

Physics

Sector:

Quantum Science; Pharmaceuticals; Artificial Intelligence

University:

University of Toronto

Program: