Quantum Computer Predicts Protein Hydration Sites with 123 Qubits, Matching Classical Precision for Drug Discovery
Summarize this article with:
Predicting how water molecules interact within protein pockets represents a critical challenge in drug discovery, influencing how effectively potential medications bind to target proteins. Daniele Loco from Qubit Pharmaceuticals, Kisa Barkemeyer and Andre R. R. Carvalho from Q-CTRL, along with Jean-Philip Piquemal from Sorbonne Universit ́e, now demonstrate a practical approach to this problem using a quantum computer.
The team successfully predicts hydration sites within protein pockets, matching the precision of established classical methods, and achieves this using hardware experiments involving up to 123 qubits. This work represents a significant step towards realising the potential of noisy intermediate-scale quantum computers for real-world applications in drug development, and suggests that quantum computation can offer advantages over classical techniques for particularly complex systems, ultimately assisting in the optimisation of drug candidates. Modeling Solvation with 3D Reference Interaction Site Models Scientists employ molecular modeling and simulation techniques to understand biomolecular behavior in solution, focusing on solvation, the interaction between a molecule and surrounding water. Accurate solvation modeling is essential for understanding protein structure, stability, and function. Researchers utilize three-dimensional Reference Interaction Site Models (3D-RISM) to calculate water density around proteins, alongside force fields such as Amber’s ff14SB and ff99SB, forming the foundation for simulating complex biological systems and predicting molecular interactions. Alongside classical methods, scientists are exploring quantum computing to enhance molecular modeling. Quantum algorithms, such as the Quantum Approximate Optimization Algorithm (QAOA) and quantum annealing, offer new approaches to optimization problems inherent in these simulations. While challenges like barren plateaus exist, researchers aim to demonstrate quantum advantage, solving problems intractable for classical computers. The accuracy of these simulations depends heavily on chosen force fields and available computational resources. Researchers continually refine parameters and develop new algorithms to reduce computational cost and improve reliability, extending this work to applications including drug discovery, where understanding ligand binding and protein structure is paramount. By addressing scalability, error mitigation, and noise, scientists strive to unlock the full potential of both classical and quantum computing for advancing our understanding of biological systems.
Quantum Computing Predicts Protein Hydration Sites Scientists have developed a novel methodology to predict the location of water molecules around proteins, a critical step in computer-aided drug discovery. This approach combines classical and quantum computation, formulating the water placement problem as a Quadratic Unconstrained Binary Optimization (QUBO). Researchers integrate a classical three-dimensional Reference Interaction Site Model (3D-RISM), which accurately describes water density, with a quantum algorithm implemented on digital quantum hardware. This allows them to map the hydration-site prediction problem onto an Ising Hamiltonian suitable for quantum computation. Experiments utilizing quantum hardware with up to 123 qubits demonstrate the ability to explore large-scale instances of this problem, successfully reproducing experimental predictions on real-life protein-ligand complexes, achieving precision comparable to classical approaches. Detailed resource estimation analysis reveals that accuracy systematically improves with an increasing number of qubits, indicating the potential for substantial gains as quantum devices scale. By creating an end-to-end workflow, the researchers advance the practical utility of quantum computing for real-world applications in the pharmaceutical industry, offering a pathway to accelerate drug discovery by improving the accuracy and efficiency of predicting protein hydration sites.
Quantum Protein Hydration Prediction with Hybrid Algorithms Scientists have achieved a breakthrough in predicting protein hydration sites using quantum computing, demonstrating the practical utility of Noisy Intermediate-Scale Quantum (NISQ) hardware for complex tasks in computer-aided drug discovery. The research team formulated the water placement problem as a Quadratic Unconstrained Binary Optimization (QUBO) and coupled a classical three-dimensional Reference Interaction Site Model (3D-RISM) with an efficient quantum optimization solver. Experiments were conducted on IBM’s Heron devices, utilizing up to 123 qubits to model realistic 3D systems without approximations. The results demonstrate that the quantum approach reproduces experimental predictions on real-life protein-ligand complexes, matching the precision of classical methods. This achievement was obtained by applying the methodology to proteins relevant to the pharmaceutical industry, including those complexed with FDA-approved drugs.
The team successfully implemented an end-to-end workflow encompassing problem encoding, quantum algorithm implementation, and hardware execution. Furthermore, a detailed resource estimation analysis reveals that accuracy systematically improves with an increasing number of qubits, indicating that full quantum utility is within reach as devices scale. This analysis forecasts the device specifications needed to address protein complexes at scales where classical methods struggle. The research also identifies advantageous situations where the quantum method could outperform classical optimization, particularly for systems where classical approaches fail to find optimal solutions. This work represents a key step toward leveraging quantum computing for drug lead optimization and setup of docking calculations, offering a promising pathway for accelerating drug discovery pipelines.
Quantum Protein Hydration Prediction with Hybrid Algorithms Scientists have demonstrated a hybrid quantum-classical approach to predicting hydration sites within protein pockets, a crucial step in computer-aided drug discovery. By formulating the problem as a Quadratic Unconstrained Binary Optimization (QUBO) and combining a classical three-dimensional reference-interaction site model with a quantum optimization solver, researchers successfully performed calculations on quantum hardware with up to 123 qubits. The results reproduce experimental predictions for real-life protein-ligand complexes, achieving accuracy comparable to established classical methods. Detailed resource estimation indicates that increasing the number of qubits systematically improves accuracy, suggesting that practical quantum utility is within reach for this type of problem. Importantly, the team observed instances where classical optimization methods struggled to find optimal solutions, while the quantum approach, or classical heuristics, outperformed them. Analysis suggests that a 900-variable instance, sufficient for many industry test cases, requires approximately 100,000 two-qubit gates and early forms of error correction, potentially achievable within five years. The authors acknowledge that demonstrating a definitive quantum advantage requires empirical evidence and comparison with state-of-the-art classical solvers, with feasible tests anticipated by 2028.
This research establishes a pathway for integrating quantum computing into drug discovery pipelines, specifically for preparing protein-ligand complexes for molecular dynamics simulations and docking calculations. 👉 More information 🗞 Practical protein-pocket hydration-site prediction for drug discovery on a quantum computer 🧠 ArXiv: https://arxiv.org/abs/2512.08390 Tags: