Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Quantum-Inspired Machine Learning for Molecular Docking

Runqiu Shu, Bowen Liu, Zhaoping Xiong, Xiaopeng Cui, Yunting Li, Wei Cui, Man-Hong Yung, Nan Qiao

TL;DR

The paper tackles blind molecular docking, a high-dimensional optimization problem, by introducing Quantum-Diffusion-Mapped Docking (QDMD), a framework that couples a score-based diffusion model with a quantum-inspired SB optimization. By encoding discrete pose decisions into a continuous tangent space via a learned encoder and evolving a quantum-inspired Hamiltonian, QDMD efficiently explores the conformational space and decodes back to 3D ligand conformations. Benchmarking on the PDBBind dataset shows that QDMD outperforms traditional docking methods and diffusion-based baselines, achieving higher Top-1 and Top-5 docking success rates and notably better performance on unseen ligands in high-precision regions ($\mathrm{RMSD}<1$). This integration of diffusion-score guidance with quantum-inspired optimization demonstrates a promising direction for fast, accurate, and generalizable molecular docking in drug discovery, leveraging both data-driven scoring and global optimization. Key contributions include the encoder bridging discrete and continuous spaces, the SB-based optimization of the docking objective, and empirical demonstration of improved generalization and precision.

Abstract

Molecular docking is an important tool for structure-based drug design, accelerating the efficiency of drug development. Complex and dynamic binding processes between proteins and small molecules require searching and sampling over a wide spatial range. Traditional docking by searching for possible binding sites and conformations is computationally complex and results poorly under blind docking. Quantum-inspired algorithms combining quantum properties and annealing show great advantages in solving combinatorial optimization problems. Inspired by this, we achieve an improved in blind docking by using quantum-inspired combined with gradients learned by deep learning in the encoded molecular space. Numerical simulation shows that our method outperforms traditional docking algorithms and deep learning-based algorithms over 10\%. Compared to the current state-of-the-art deep learning-based docking algorithm DiffDock, the success rate of Top-1 (RMSD<2) achieves an improvement from 33\% to 35\% in our same setup. In particular, a 6\% improvement is realized in the high-precision region(RMSD<1) on molecules data unseen in DiffDock, which demonstrates the well-generalized of our method.

Quantum-Inspired Machine Learning for Molecular Docking

TL;DR

The paper tackles blind molecular docking, a high-dimensional optimization problem, by introducing Quantum-Diffusion-Mapped Docking (QDMD), a framework that couples a score-based diffusion model with a quantum-inspired SB optimization. By encoding discrete pose decisions into a continuous tangent space via a learned encoder and evolving a quantum-inspired Hamiltonian, QDMD efficiently explores the conformational space and decodes back to 3D ligand conformations. Benchmarking on the PDBBind dataset shows that QDMD outperforms traditional docking methods and diffusion-based baselines, achieving higher Top-1 and Top-5 docking success rates and notably better performance on unseen ligands in high-precision regions (). This integration of diffusion-score guidance with quantum-inspired optimization demonstrates a promising direction for fast, accurate, and generalizable molecular docking in drug discovery, leveraging both data-driven scoring and global optimization. Key contributions include the encoder bridging discrete and continuous spaces, the SB-based optimization of the docking objective, and empirical demonstration of improved generalization and precision.

Abstract

Molecular docking is an important tool for structure-based drug design, accelerating the efficiency of drug development. Complex and dynamic binding processes between proteins and small molecules require searching and sampling over a wide spatial range. Traditional docking by searching for possible binding sites and conformations is computationally complex and results poorly under blind docking. Quantum-inspired algorithms combining quantum properties and annealing show great advantages in solving combinatorial optimization problems. Inspired by this, we achieve an improved in blind docking by using quantum-inspired combined with gradients learned by deep learning in the encoded molecular space. Numerical simulation shows that our method outperforms traditional docking algorithms and deep learning-based algorithms over 10\%. Compared to the current state-of-the-art deep learning-based docking algorithm DiffDock, the success rate of Top-1 (RMSD<2) achieves an improvement from 33\% to 35\% in our same setup. In particular, a 6\% improvement is realized in the high-precision region(RMSD<1) on molecules data unseen in DiffDock, which demonstrates the well-generalized of our method.
Paper Structure (10 sections, 10 equations, 2 figures, 4 tables, 1 algorithm)

This paper contains 10 sections, 10 equations, 2 figures, 4 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Background
  3. Molecular Docking
  4. Machine learning-based Docking
  5. Quantum-inspired Algorithm
  6. Method
  7. Numerical Simulation
  8. Benchmarking
  9. Simulation Results
  10. Conclusion

Figures (2)

  • Figure 1: Overall process of blind molecular docking using QDMD.The main steps of QDMD include random generation of initial positions and conformations, encoding and decoding processes, optimization of Insing values using quantum-inspired algorithm, ranking and filtering of results.
  • Figure 2: Cumulative RMSD histogram results for different docking methods. The top graph represents the experimental results on 363 pairs of protein-ligand complexes selected based on previous work. The bottom graph represents the experimental results on the newly released ligand dataset.