Fig. 1
Illustration of predicting DDIs for a newly given drug. In the left panel, nodes in DDI network represent drugs. The hollow nodes are known drugs (numbered from 1 to 7) and the solid lines between them denote their interactions. Blue lines are enhancive interactions and yellow lines are degressive interactions. The node in red respectively is the newly given drug, tagged as X. In the right panel, the adjacent matrix is shown. The cells in it are filled with blue, yellow and red, accounting for the types of DDIs and drug pairs of interest respectively. All the pairwise entries among {d1,d2…,d7} are used to train the model, the entries in the red cells denote the testing entries. Our problem is to determine which known drugs could interact with the new drug X and what type these potential interactions are