Companion repository for: Auditing preclinical evidence at every drug-discovery decision point; dimensional, validation-density, and coverage-gate methods demonstrated in several oncology programs

Companion repository for the manuscript: Alshanqeeti A. "Auditing preclinical evidence at every drug-discovery decision point: dimensional, validation-density, and coverage-gate methods demonstrated in several oncology programs." The framework is a three-component decision rule (redundancy gate via participation ratio and PC2 variance share, validation-density gate via VDI ratio, coverage gate via partial-state vector) that operates on a generic input structure: a rows-by-features preclinical evidence matrix paired with proposed indication labels. The methodology applies at every drug-discovery decision point where commitment is being made on the basis of preclinical evidence. Empirical demonstration in the manuscript covers two decision points: Phase III commitment (nine oncology drug programs) and target identification (FLT3 inhibitor program). The framework is input-data-agnostic and characterizes the structural state of whatever input is supplied, against transparent prespecified thresholds. This deposit contains curated data, analysis scripts, intermediate results, figures, and a standalone audit protocol. All scripts run under Python 3 with pandas, numpy, scipy, and matplotlib. The reproducibility_verification directory contains a script that runs each pipeline from scratch and confirms key summary values match the reported counts. License: Creative Commons Attribution 4.0 International (CC BY 4.0) for data and documents; MIT License for code. Citation: Alshanqeeti A. (2026). Companion repository for "Auditing preclinical evidence at every drug-discovery decision point." Zenodo. https://doi.org/10.5281/zenodo.20089005