Optimizing Quantum Data Embeddings for Ligand-Based Virtual Screening

TL;DR

This paper explores quantum-classical hybrid embedding strategies to enhance molecular representations for ligand-based virtual screening, demonstrating improved performance over classical methods, especially with limited data.

Contribution

It introduces novel quantum-classical hybrid embedding approaches and evaluates their effectiveness across benchmark datasets, showing their potential in drug discovery tasks.

Findings

Quantum and hybrid embeddings outperform classical baselines.

Hybrid methods are especially effective in limited-data scenarios.

Results suggest quantum embeddings can be data-efficient tools for virtual screening.

Abstract

Effective molecular representations are essential for ligand-based virtual screening. We investigate how quantum data embedding strategies can improve this task by developing and evaluating a family of quantum-classical hybrid embedding approaches. These approaches combine classical neural networks with parameterized quantum circuits in different ways to generate expressive molecular representations and are assessed across two benchmark datasets of different sizes: the LIT-PCBA and COVID-19 collections. Across multiple biological targets and class-imbalance settings, several quantum and hybrid embedding variants consistently outperform classical baselines, especially in limited-data regimes. These results highlight the potential of optimized quantum data embeddings as data-efficient tools for ligand-based virtual screening.