Quantum descriptor–structure fusion for DFT-level frontier orbital prediction

This repository contains the data and supporting files associated with the study “Quantum Descriptor–Structure Fusion for DFT-Level Frontier Orbital Prediction.” The study presents a machine-learning framework that combines semiempirical PM7 frontier-orbital descriptors with RDKit molecular descriptors and radius-2 Morgan fingerprints to predict DFT-level HOMO energy, LUMO energy, and HOMO–LUMO gap. The deposited materials include the processed PM7–DFT matched dataset, molecule identifiers for the training, validation, and held-out TestFinal sets, generated molecular descriptors and fingerprints, model predictions, residual-analysis tables, and the source data underlying the figures and tables. The repository also includes the processed Gaussian-derived HOMO–LUMO gap and dipole-moment data used in the erlotinib–nanocarrier electronic-response case study. The original DFT frontier-orbital dataset was reported by Pereira and co-workers and is publicly available through Figshare at https://doi.org/10.6084/m9.figshare.3384184. Users of this repository should also cite the original dataset and its associated publication.