Bindwell/PLBA

这是一个全面的蛋白质-配体结合亲和力数据集，从多个来源收集而成。数据集包含多个分割，每个分割对应于特定的数据源，如glaser、pdbbind-2013-core、davis等。训练集使用Glaser数据集，验证集使用PDBBind 2013 Core数据集。数据集的列包括蛋白质序列、配体的标准SMILES表示、结合亲和力（微摩尔浓度）、结合亲和力的负对数（摩尔浓度）、归一化的结合亲和力、重复蛋白质-配体对的平均结合亲和力和标准差。

This is a comprehensive collection of protein-ligand binding affinity data compiled from multiple sources. The dataset includes multiple splits, each corresponding to a specific source such as glaser, pdbbind-2013-core, davis, etc. The training set uses the Glaser dataset, and the validation set uses the PDBBind 2013 Core dataset. The columns of the dataset include the protein sequence, the canonical SMILES representation of the ligand, the binding affinity in micromolar concentration, the negative logarithm (base 10) of the affinity in molar concentration, the normalized binding affinity, the mean binding affinity for duplicate protein-ligand pairs, and the standard deviation of the binding affinity for duplicate protein-ligand pairs.