v1.0.1

Overview

Documentation and research-record update following the initial v1.0.0 release.

Updated

• Refined README methodology and experimental-scope descriptions
• Added the complete Graph Neural Network methodology and evaluation section
• Clarified the distinction between the original classical benchmark and the matched classical-vs-GNN audit
• Updated reported benchmark and GNN results
• Clarified PiGNN as a future research direction rather than a completed model
• Updated CHANGELOG and research provenance documentation
• Corrected GitHub Pages documentation links and figure paths
• Improved repository documentation and reproducibility references

Research record

This release does not replace the historical v1.0.0 record. It provides a subsequent version of the repository with corrected and expanded documentation.

The v1.0.0 archival record remains available through its original DOI.

Full Changelog: updated-report...v1.0.1

v1.0.0

AChE Molecular Machine Learning — v1.0.0

First formal research release of the AChE molecular machine-learning project.

This release contains the documented source code, experimental artifacts,
diagnostic outputs, forensic audit, and complete scientific report associated
with the current AChE pIC50 modelling workflow.

Included

• ChEMBL AChE activity-data pipeline
• Molecular curation and standardization
• IC50 → pIC50 processing
• Morgan fingerprints
• RDKit molecular descriptors
• Scaffold-aware evaluation
• Random Forest
• XGBoost
• LightGBM
• Optuna optimization
• Grouped cross-validation
• Held-out evaluation
• Graph Neural Network evaluation
• Diagnostic analyses
• Forensic audit artifacts
• Complete scientific report

Scientific note

This release is an evidence-bound research snapshot. The included forensic
audit documents methodological limitations identified during development,
including a unit-consistency issue concerning potency thresholding.

The release does not claim state-of-the-art performance or publication-ready
benchmark validity.

Citation

Please cite this software release using the repository's CITATION.cff.

A permanent archival DOI will be added after Zenodo deposition.

DOI: 10.5281/zenodo.23182018

Full Changelog: https://github.com/TheHashiramaSenju/AcHe-QSAR-modelling/commits/report