Data and Software for "Conformer-Resolved Optical Rotation Prediction from Three-Dimensional Quantum and Semiempirical Molecular Fields Using 3D Convolutional Neural Networks"

Research data and software supporting the manuscript “Conformer-Resolved Optical Rotation Prediction from Three-Dimensional Quantum and Semiempirical Molecular Fields Using 3D Convolutional Neural Networks.” The archive contains the final 82,790-record systematic conformer dataset and associated Cartesian structures and conformer-resolved sTD-DFT optical-rotation targets; the independent 705-conformer CREST application benchmark for 65 molecular identities; the 15 full-dataset deployment models used for the reported benchmark; computational-environment information; and a snapshot of the associated analysis and machine-learning code. Large precomputed volumetric descriptor tensors are not included because they occupy several hundred gigabytes and can be regenerated from the deposited structures using the provided scripts and computational settings.