Fig. 3
From: Structure-based, deep-learning models for protein-ligand binding affinity prediction
The grid representation and learning architecture of \(T_{Grid-CNN}\) models. A One grid that covers both protein and ligand atoms. B Two concatenated grids that featurize protein and ligand atoms separately. C An augmented grid representation. D The learning architecture for PLBAP (used by KDEEP). The red numbers indicate the numbers of filters in convolution layers or the numbers of units in dense layers