Fig. 1

SAveRUNNER conceptual organization. SAveRUNNER takes as input a list of drug-target interactions and disease-gene associations, and releases as output predicted drug-disease associations by performing seven steps (dashed box of this flowchart). In particular, Steps 1–3 bring to the construction of a proximity-based bipartite drug-disease network, where nodes are both drugs and diseases, edges are the statistically significant drug-disease associations (p value \(\le 0.05\), or z-score \(\le -1.65\)), weighted according to the proximity values; Steps 4–7 bring to the construction of a similarity-based bipartite drug-disease network, where the weights represent the adjusted similarity measure computed to prioritize the predicted drug-disease associations by rewarding the associations between drugs and diseases belonging to the same network neighborhood. Finally, the drug-disease associations predicted by SAveRUNNER were evaluated by performing a ROC curve probability analysis (solid line box of this flowchart). The ROC curve is computed for SAveRUNNER algorithm by plotting the true positive rate (TPR) placed on Y-axis against the false positive rate (FPR) placed on X-axis at various threshold settings. Diagonal grey line represents the line of no-discrimination between positive class (known drug-disease associations) and negative class (unknown drug-disease associations)