ADMET Profiling of a Mitochondria-Focused Compound Library: Physicochemical Baselines, Membrane Permeability Classification, and Tier-Based Filtering for AI-Driven Drug Discovery

Data deposit accompanying the manuscript: ADMET Profiling of a Mitochondria-Focused Compound Library: Physicochemical Baselines, Membrane Permeability Classification, and Tier-Based Filtering for AI-Driven Drug Discovery This dataset contains ADMET baseline profiles for 201 mitochondrial therapeutic compounds across 25 targets, including physicochemical property distributions (molecular weight, LogP, TPSA, HBD, HBA, rotatable bonds), membrane permeability classifications, mitochondria-specific anomaly identification (TPP+ conjugate misclassification), and comparative benchmarking against published ADMET prediction tools (SwissADME, pkCSM, ADMETlab). Constraint thresholds derived from empirical distributions are provided as numerical values for reproducibility. Patent pending: US Provisional Application 64/018,624, filed March 27, 2026. This deposit contains ADMET baseline data only. Manuscript series: DrugSynth AI / MitoCorex — Computational Drug Discovery for Mitochondrial Diseases (10 manuscripts).随稿件一同提交的数据集： 聚焦线粒体的化合物库ADMET（吸收、分布、代谢、排泄、毒性）表征：面向AI驱动药物发现的理化基线、膜通透性分类及层级过滤策略 本数据集涵盖201种线粒体治疗性化合物针对25个靶点的ADMET基线表征数据，包括理化性质分布（分子量、LogP、拓扑极性表面积（TPSA, Topological Polar Surface Area）、氢键供体（HBD, Hydrogen Bond Donor）、氢键受体（HBA, Hydrogen Bond Acceptor）、可旋转键数）、膜通透性分类、线粒体特异性异常识别（TPP+缀合物错分类情况），以及与已发表ADMET预测工具（SwissADME、pkCSM、ADMETlab）的对比基准测试结果。研究同时提供基于经验分布推导得到的约束阈值数值，以保障研究可重复性。 专利待审：美国临时专利申请第64/018,624号，于2026年3月27日提交。本数据集仅包含ADMET基线数据。 稿件系列：DrugSynth AI / MitoCorex — 面向线粒体疾病的计算药物发现（共10篇稿件）