Fig 1.
The workflow of the identification of significant MeSH terms.
Table 1.
Identification of significant DDI-related compounds and proteins from MeSH terms.
Fig 2.
ROC curves for predicting DDI-related drug terms for cyclosporine, rifampin and theophylline.
Fig 3.
ROC curves for predicting DDI-related protein terms for cyclosporine, rifampin and theophylline.
Fig 4.
Protein-drug co-occurrence heatmaps for cyclosporine, rifampin and theophylline.
The rows and columns represent drug and protein terms respectively, and each cell is the normalized count of co-occurrences of the two terms. The normalized count can be calculated as normalized count = count/ total row count.
Fig 5.
Phenomena-drug co-occurrence heatmaps for cyclosporine, rifampin and theophylline.
The rows and columns represent drug and phenomena terms respectively, and each cell is the normalized count of co-occurrences of the two terms. The normalized count can be calculated as normalized count = count/ total row count.
Fig 6.
The social network of drug, proteins, phenomena and DDI types for three drug pairs, (A) cyclosporine-itraconazole, (B) rifampin-quindine, and (C) theophylline-omeprazole.
Drugs, proteins, phenomena and DDI types are shown in red, blue, green and orange, respectively.