Network Crosstalk as a Basis for Drug Repurposing

Abstract

The need for systematic drug repurposing has seen a steady increase over the past decade and may be particularly valuable to quickly remedy unexpected pandemics. The abundance of functional interaction data has allowed mapping of substantial parts of the human interactome modeled using functional association networks, favoring network-based drug repurposing. Network crosstalk-based approaches have never been tested for drug repurposing despite their success in the related and more mature field of pathway enrichment analysis. We have, therefore, evaluated the top performing crosstalk-based approaches for drug repurposing. Additionally, the volume of new interaction data as well as more sophisticated network integration approaches compelled us to construct a new benchmark for performance assessment of network-based drug repurposing tools, which we used to compare network crosstalk-based methods with a state-of-the-art technique. We find that network crosstalk-based drug repurposing is able to rival the state-of-the-art method and in some cases outperform it.

Keywords: benchmark; drug repositioning; drug repurposing; functional association network; network crosstalk; network-based; shortest path.

FIGURE 1

Network-based drug repurposing, proximity vs. crosstalk, for a given disease and drug. In crosstalk, the number of network links between all drug targets and disease genes is used while, in proximity, the length of the single shortest path between any drug target and any closest disease gene is used.

FIGURE 2

The proportion of gene overlap (marked “TRUE”) between drug targets and disease genes in the different sets of positive and negative drug–disease combinations. The proportion of not overlapping genes is marked “FALSE”.

FIGURE 3

Drug–disease association–based performance on the Guney2016 benchmark. Performance of the different drug repurposing tools: ANUBIX (green), BinoX (orange), NEAT (purple), and proximity (prox, pink) on (A) ROC curves, where the dotted line represents random prediction, and (B) AUROC using sampled sets from the benchmark containing equal numbers of positive and negative drug–disease combinations. The pairwise Wilcoxon rank sum test was used to assess the significance of difference on AUROC. FDR-corrected p-values were obtained using the Benjamini–Hochberg procedure.

FIGURE 4

Drug–disease association–based performance on the time-stamped benchmark. Performance of the different drug repurposing tools: ANUBIX (green), BinoX (orange), NEAT (purple), and proximity (prox, pink) on (A) ROC curves, where the dotted line represents random prediction, and (B) AUROC using sampled sets from the benchmark containing equal numbers of positive and negative drug–disease combinations. The pairwise Wilcoxon rank sum test was used to assess the significance of difference on AUROC. FDR-corrected p-values were obtained using the Benjamini–Hochberg procedure.

FIGURE 5

Drug–disease association–based performance on the FCbench. Performance of the different drug repurposing tools: ANUBIX (green), BinoX (orange), NEAT (purple), and proximity (prox, pink) on (A) ROC curves, where the dotted line represents random prediction, and (B) AUROC using sampled sets from the benchmark containing equal numbers of positive and negative drug–disease combinations. The pairwise Wilcoxon rank sum test was used to assess the significance of difference on AUROC. FDR-corrected p-values were obtained using the Benjamini–Hochberg procedure.

FIGURE 6

Drug–drug similarity–based performance on the Guney2016 benchmark. Performance of the different drug repurposing tools: ANUBIX (green), BinoX (orange), NEAT (purple), and proximity (prox, pink) on (A) ROC curves, where the dotted line represents random prediction, and (B) AUROC using sampled sets from the benchmark containing equal numbers of positive and negative drug–disease combinations. The pairwise Wilcoxon rank sum test was used to assess the significance of difference on AUROC. FDR-corrected p-values were obtained using the Benjamini–Hochberg procedure.

FIGURE 7

Drug–drug similarity–based performance on the time-stamped benchmark. Performance of the different drug repurposing tools: ANUBIX (green), BinoX (orange), NEAT (purple), and proximity (prox, pink) on (A) ROC curves, where the dotted line represents random prediction, and (B) AUROC using sampled sets from the benchmark containing equal numbers of positive and negative drug–disease combinations. The pairwise Wilcoxon rank sum test was used to assess the significance of difference on AUROC. FDR-corrected p-values were obtained using the Benjamini–Hochberg procedure.

FIGURE 8

Drug–drug similarity–based performance on the FCbench. Performance of the different drug repurposing tools: ANUBIX (green), BinoX (orange), NEAT (purple), and proximity (prox, pink) on (A) ROC curves, where the dotted line represents random prediction, and (B) AUROC using sampled sets from the benchmark containing equal numbers of positive and negative drug–disease combinations. The pairwise Wilcoxon rank sum test was used to assess the significance of difference on AUROC. FDR-corrected p-values were obtained using the Benjamini–Hochberg procedure.