A Point Cloud-Based Deep Learning Model for Protein Docking Decoys Evaluation

Protein-protein docking reveals the process and product in protein interactions. Typically, a works with model sampling, then an evaluation method is used to rank near-native models out from large pool of generated decoys. In practice, stage bottleneck perform accurate docking. this paper, PointNet, deep learning algorithm based on point cloud, applied evaluate models. The proposed architecture able directly learn representations carrying geometrical properties atomic attributes 3D structural data experimental results show that informative can benefit our outperform other algorithms.