Ligand binding affinity prediction with fusion of graph neural networks and 3D structure-based complex graph

We proposed a novel deep fusion graph neural networks framework named FGNN to learn the protein–ligand interactions from the 3D structures of protein–ligand complexes.