PatSnap/chemical-structure

Curated Chemical Structures & Properties数据集是一个包含5,000种化合物的资源，用于化学信息学和AI驱动的药物发现。它提供了标准化的分子表示（如Canonical SMILES, Isometric SMILES, InChI, InChIKey）以及计算出的物理化学性质和层次化的ClassyFire分类。数据集适用于训练分子生成模型、预测ADMET性质或构建结构-活性关系（SAR）模型。关键特征包括标准化的分子表示、丰富的物理化学特征和深层次的分类注释。

The Curated Chemical Structures & Properties dataset is a 5,000-compound resource for cheminformatics and AI-driven drug discovery. It provides standardized molecular representations (Canonical SMILES, Isometric SMILES, InChI, InChIKey) alongside computed physicochemical properties and hierarchical ClassyFire taxonomic classifications. Ideal for training molecular generative models, predicting ADMET properties, or building structure-activity relationship (SAR) models. Key features include standardized representations, rich physicochemical features, and deep taxonomic classification.