Skip to main content
. 2023 Dec 22;40(1):btad774. doi: 10.1093/bioinformatics/btad774

Figure 1.

Figure 1.

Schematic of GeNNius architecture. GeNNius inputs a graph containing drug (red) and protein (blue) nodes, where Nd and Np represent the number of drugs and proteins, respectively. First, a GNN generates node representations with an embedding of dimension d =17. Second, a NN-based classifier aims at learning the existence of an edge given a set of concatenations of drug and protein embeddings. Note that at this stage a negative set of edges is generated.