Data and Code for "Advancing QSPR with Sum-Connectivity Descriptors: Physicochemical and Antibacterial Modelling via Molecular Graph Connectivity"

This Zenodo record provides the reproducibility package associated with the article “Advancing QSPR with Sum-Connectivity Descriptors: Physicochemical and Antibacterial Modelling via Molecular Graph Connectivity”. The package contains the curated antibacterial Escherichia coli modelling dataset derived from publicly available ChEMBL bioactivity records, together with Python scripts and supporting documentation used to reproduce the QSPR/QSAR analyses reported in the article. The fixed dataset contains 6,657 unique compounds and 134 variables, including ChEMBL identifiers, assay information, physicochemical properties, RDKit molecular descriptors, classical degree-based topological indices, proposed sum-connectivity indices, and the antibacterial activity endpoint pMIC. The MIC values were standardized and transformed into pMIC values for antibacterial activity modelling. The archived materials include: (i) the fixed merged modelling dataset, dataset_FIXED_full.csv; (ii) RDKit descriptor variables; (iii) classical degree-based topological indices; (iv) proposed sum-connectivity descriptor analogues; (v) portable Python scripts for model fitting and validation; (vi) original Python scripts used in the article workflow; (vii) repeated cross-validation and Murcko scaffold validation workflows; (viii) y-randomization and permutation-importance analysis code; and (ix) supporting documentation, including README, data dictionary, Zenodo upload checklist, requirements file, conda environment file, citation file, checksum file, and license files. The package supports reproducibility of the QSPR benchmarks, QSAR pMIC prediction, comparison between classical and sum-connectivity descriptor families, and diagnostic analyses reported in the article. The original bioactivity records remain publicly available from the ChEMBL database.