Performance Evaluation of Quantum Machine Learning Models for Drug Target Identification

Abstract

Drug target identification is a crucial step in the drug discovery pipeline. With the increasing availability of biological and chemical datasets, machine learning techniques have shown great promise in predicting potential drug targets. Recent advances in quantum computing have opened up new possibilities of applying quantum machine learning algorithms to computational drug discovery. In this work, we benchmark the performance of various classical and quantum machine learning models on drug target prediction tasks. We train supervised classification models on benchmark datasets of chemical compounds labeled with their target protein. We compare quantum classifiers implemented using variational quantum circuits against classical neural networks and kernel methods. Our results demonstrate that certain quantum models can achieve significantly higher accuracy than classical approaches in identifying drug targets across various protein target families. The quantum advantage is more pronounced on datasets with greater molecular diversity. Our work provides useful insights into the practical value of quantum machine learning for an important real-world application in computational biology. The performance evaluations presented serve as a guide for applying quantum algorithms to develop more effectively in silico drug discovery pipelines.