Quantum Approximate Optimization Algorithms for Molecular Docking

TL;DR

This paper presents a Digitized Counterdiabatic QAOA (DC-QAOA) approach to molecular docking, and concludes that binding interactions represent the anticipated exact solution.

Abstract

Molecular docking is a critical process for drug discovery and challenging due to the complexity and size of biomolecular systems, where the optimal binding configuration of a drug to a target protein is determined. Hybrid classical-quantum computing techniques offer a novel approach to address these challenges. The Quantum Approximate Optimization Algorithm (QAOA) and its variations are hybrid classical-quantum techniques, and a promising tool for combinatorial optimization challenges. This paper presents a Digitized Counterdiabatic QAOA (DC-QAOA) approach to molecular docking. Simulated quantum runs were conducted on a GPU cluster. We examined 14 and 17 nodes instances - to the best of our knowledge the biggest published instance is 12-node at Ding et al. and we present the results. Based on computational results, we conclude that binding interactions represent the anticipated exact solution. Additionally, as the size of the examined instance increases, the computational times exhibit a significant escalation.

Figures (6)

• Figure 1: Illustration of converting molecular docking to clique finding problem. a. Crystal structure of protein-ligand binding complex ding2024molecular
• Figure 2: A. Cost evolution comparison of applying a warm-start technique and not. The warm-start technique provides a better initial convergence point for the optimization process. For this instance example the convergence time is almost the same. B. Histogram of the top 10 sampled frequencies for 14-qubit instance. The most sampled bitstring correspond to the actual ground-truth state. C. Graph of 14-qubit instance with the solution colored.
• Figure 3: In Panel (a), (b) and (c) we illustrate the cost evolution determined by DC-QAOA approach by using COBYQA with linear warm-start, quadratic warm-start and cold-start. For each case, we generated and optimized with varying penalties ( $P=1$, $P=2$), varying layers ($n = 1$, $n = 2$) and 40000 iterations.
• Figure 4: Different method runs (quadratic, linear and cold-start), compared to each other, with varying layers and penalties. All of these are COBYQA optimization runs.
• Figure 5: Sampled frequencies by bitstrings for 17-qubits with solution of pharmacophores. We are selecting the top 10 of each experiment run, where each bitstring with the highest cost, represents the solution.