A Quantum Framework for Protein Binding-Site Structure Prediction on Utility-Level Quantum Processors.

Area of Science:

• Quantum Computing
• Structural Biology
• Computational Chemistry

Background:

• Accurate protein active-site structure prediction is crucial but challenging for short, flexible peptide fragments.
• Conventional and simulation-based methods often struggle with these complex structures.

Purpose of the Study:

• To develop a quantum computing framework for predicting protein active-site structures on utility-level quantum processors.
• To address limitations of existing methods for small, flexible peptide fragments.

Main Methods:

• Utilized the Variational Quantum Eigensolver (VQE) to frame structure prediction as a ground-state energy minimization problem.
• Encoded amino acid connectivity and constraints (steric, geometric, chirality) into a problem-specific Hamiltonian using sparse Pauli operators.
• Implemented a two-stage architecture for energy estimation and noise-mitigated measurement decoding on a quantum processor.

Main Results:

• The quantum framework was evaluated on 30 protein fragments, including therapeutic targets, using the IBM-Cleveland Clinic quantum processor.
• Benchmarked against AlphaFold3 (AF3) and classical simulations, the quantum approach demonstrated superior performance in Root-Mean-Square Deviation (RMSD).
• The quantum method also showed improved docking efficacy compared to AF3 and classical baselines.

Conclusions:

• The developed quantum computing framework offers a practical end-to-end pipeline for biologically relevant protein structure prediction on real quantum hardware.
• The study highlights the engineering feasibility and potential of near-term quantum devices for complex structural biology challenges.
• This quantum approach surpasses current state-of-the-art methods in accuracy and efficacy for specific protein fragment prediction tasks.