A network data is composed of nodes and edges. An example of such network data would be social network data where nodes are people and their interests and edges are interconnections between them[2-5]. Many useful applications such as customized suggestions for social media users have been developed through the use of Machine/Deep learning algorithms which accomplish this through node classification and link prediction protocols.
The bindingDB database was downloaded and a network was constructed using NetworkX wherein the nodes where compounds and proteins and edges where the interactions between them. Lower the IC50 value for a compound to inhibit a particular protein, the shorter the edges were that link them together. Each compound is identified using the PubChem Compound ID (CID) and proteins are identified with the Protein Data Bank ID (PDB ID).To generate 2D embeddings of the network, the node2vec [29] python package was used. The module learnt the embeddings of 65 graphs and they were used to perform a machine learning/deep learning based multi-class classification .