NutriDock-CRC processed dataset and molecular features for exposure-informed colorectal alteration screening

This repository contains the processed data resources used in the NutriDock-CRC framework, a subject-level cheminformatics pipeline for colorectal alteration screening. The package integrates individualized dietary exposure data with docking-derived molecular features and graph-based molecular resources. It includes: (i) a subject-by-chemical exposure table derived from dietary assessment, where each row corresponds to one participant and each chemical column represents estimated intake for a retained food-related compound; (ii) a curated molecular feature table indexed by compound CID and protein target, containing docking affinities, SMILES-derived physicochemical descriptors, and compact voxel-based interaction summaries extracted from ligand–protein complexes; and (iii) a compressed collection of molecular graph resources used for exploratory graph-based representations of ligands, including serialized graph objects and text exports describing node and edge features. These files are intended to support reproducibility of the processed feature-construction stage of the framework. The dataset represents de-identified, processed data products rather than raw clinical records. In the associated study, the integrated representation combines dietary burden, aggregated ligand-level descriptors, and protein-specific interaction summaries to build subject-level predictors for Normal, Polyp, and Neoplasia/Altered colorectal profiles.