Q-CaDD: accelerating in silico methodologies with quantum computation and machine learning for Epidermal growth factor receptor

The Epidermal Growth Factor Receptor (EGFR) is a transmembrane protein in the Receptor Tyrosine Kinases (RTKs) family, playing a central role in cell growth, motility, and differentiation. Mutations in EGFR are associated with various cancers, notably Non-Small Cell Lung Cancer (NSCLC), due to overexpression and dysregulated signaling. To support early-stage drug discovery targeting EGFR, we introduce Q-CaDD (Quantum-enhanced Computer-aided Drug Design), a hybrid quantum–classical computational framework. Q-CaDD integrates quantum machine learning (QML), classical models, molecular docking, and multi-stage ligand filtering. Over 200,000 ligands were generated using STONED-SELFIES and filtered using Ghose and Quantitative Estimate of Drug-Likeness (QED) criteria. Docking simulations were performed using AutoDock Vina, and toxicity prediction employed an ensemble of machine learning models including a Quantum Support Vector Machine (QSVM). Model evaluation was conducted on the Tox21 dataset using the NR-AR (Androgen Receptor) pathway as a proof-of-concept toxicity endpoint. The hybrid ensemble achieved an AUC-ROC of 0.83, demonstrating improved generalisation relative to standalone classical models under the same evaluation setting and competitive performance with respect to reported benchmark results. Several compounds exhibited favourable predicted affinity and toxicity profiles and are identified as preliminary candidate molecules pending further experimental validation. Overall, this study demonstrates that quantum-enhanced components can contribute complementary predictive signals within hybrid drug discovery workflows, even under the constraints of current Noisy Intermediate-Scale Quantum (NISQ) hardware.