Comparison of residue mapping strategies.

We applied five residue mapping algorithms to three grids of predicted pockets (ConCavityL, ConCavityP, ConCavityS ). This table lists the PR-AUC for identifying ligand binding residues in the LigASite apo dataset for each combination. Our Blur algorithm achieves the best performance for each grid type.

我们将五种残基映射算法应用于三组预测口袋网格（ConCavityL、ConCavityP、ConCavityS）。本表格列出了在LigASite空态数据集（LigASite apo dataset）中，针对每种算法与网格的组合识别配体结合残基的PR-AUC（精确召回曲线下面积，Precision-Recall Area Under the Curve）值。我们的Blur算法在所有网格类型下均取得了最优性能。