Fig. 2.

Knowledge graph build using two homogeneous similarity matrices and DTI data where the first-hop neighbor of the target protein includes the protein with the same structure from the dataset, the second-hop neighbor contains molecules from the dataset and the third hop contains the generated molecules