Fig 1.
A micro-to-macro drug-centric heterogeneous network DSMN construction.
An example of segmenting drug SMILES using the BRICS [2] rules to construct the drug-substructure network, and the heterogeneous network DSMN is reconstructed by assembling all drug-related networks.
Fig 2.
The sub-figure (a) depicts all the motifs presented in our work. DRSS(Ij), DISS(Ip), DRSM (Ii), DISM(Iu) denote the four motif-driven hypergraphs constructed on drug related and have the same substructure, drug independent and have the same substructure, drug related and have the same molecular interactions, and drug independent and have the same molecular interactions motifs groups, respectively. The sub-figure (b) is a real example of driving the hypergraph base on triangular motif. It shows that the hypergraph method will make the relation between some nodes with high-order relations closer. The sub-figure (c) draws the process of multi-branches hypergraph attention and graph convolutional networks inferring new drug-related interactions.
Fig 3.
Prediction results of HGDrug and tasks-specific baseline models on the four tasks.
Fig 4.
Ablation experiments explore the contribution of neural network framework and different category hypergraph branches.
Table 1.
Statistical information for the four interactions datasets: The number of drugs, drug-related nodes, fragments and interaction pairs.
Table 2.
Prediction results of HGDrug and general baselines models on 4 drug-interactions interactions datasets.
The best performance is marked in bold and the second best is underlined to facilitate reading.
Table 3.
Top 20 novel DDIs predictions and their validation.
If the DrugBank drug interaction records the interaction information between the two drugs, the “label” is set as 1, otherwise 0. The Table E in S1 Text describes the details of DDIs.
Fig 5.
Network visualization and analysis of the drug-disease associations predicted by HGDrug.
(a) In the network, the predicted novel top 100 DDiI network is visualized. The pink nodes are diseases and the green nodes are drugs. The label of the node represents the ID of the drugs (Drugbank_ID) or diseases (UMLS_ID). The node size denotes the degree. The weight of edges (drug-disease associations) denotes the predicted score by HGDrug. (b) Analysis of top 5 drug-disease associations predicted by HGDrug. Four drugs are included in the top 5 predicted potential associations. The similarity between drugs comes from the drug features learned by HGDrug in the DDiI task, and the drug most similar to these four drugs are selected respectively. This network was generated by Gephi (https://gephi.org).
Fig 6.
The 4 drugs and their corresponding top 1 drug structures from Fig 5(b).
The same substructures of drug pairs are marked in orange.
Fig 7.
Analysis of the predicted results for the four tasks of paclitaxel.
The network contains the known interactions of paclitaxel and the most relevant entity nodes predicted by HGDrug for paclitaxel in the four tasks. The known related drugs of four nodes are given for visualizing the drugs that these nodes share with paclitaxel.
Table 4.
Computation of motif-induced adjacency matrices.