A Comparison of Network-Based Methods for Drug Repurposing along with an Application to Human Complex Diseases

Abstract

Drug repurposing strategy, proposing a therapeutic switching of already approved drugs with known medical indications to new therapeutic purposes, has been considered as an efficient approach to unveil novel drug candidates with new pharmacological activities, significantly reducing the cost and shortening the time of de novo drug discovery. Meaningful computational approaches for drug repurposing exploit the principles of the emerging field of Network Medicine, according to which human diseases can be interpreted as local perturbations of the human interactome network, where the molecular determinants of each disease (disease genes) are not randomly scattered, but co-localized in highly interconnected subnetworks (disease modules), whose perturbation is linked to the pathophenotype manifestation. By interpreting drug effects as local perturbations of the interactome, for a drug to be on-target effective against a specific disease or to cause off-target adverse effects, its targets should be in the nearby of disease-associated genes. Here, we used the network-based proximity measure to compute the distance between the drug module and the disease module in the human interactome by exploiting five different metrics (minimum, maximum, mean, median, mode), with the aim to compare different frameworks for highlighting putative repurposable drugs to treat complex human diseases, including malignant breast and prostate neoplasms, schizophrenia, and liver cirrhosis. Whilst the standard metric (that is the minimum) for the network-based proximity remained a valid tool for efficiently screening off-label drugs, we observed that the other implemented metrics specifically predicted further interesting drug candidates worthy of investigation for yielding a potentially significant clinical benefit.

Keywords: drug repurposing; network medicine; network theory.

Figure 6

Network-based proximity measures. Schematic representation of the proximity measures computed between target proteins t of drug module T and disease genes s of disease module S according to five different metrics (a–e) described by Equations (1)–(5).

Figure 1

Workflow of the analysis. Input data are the human interactome network, the disease-gene associations from DisGeNET and the drug-targets interactions from DrugBank. The proximity measure between drug-targets and disease genes is computed by using five different metrics, including the standard minimum and the other here-proposed ones (i.e., maximum, mean, median, mode). The resulting candidate drugs are then compared among each metric, and metric-specific drugs are then discussed.

Figure 2

Number of off-label predicted drugs according the five different metrics for each analyzed disease. (a) The table reports, for each disease and each metric, the total number of predicted drugs, the number of predicted drugs that already have a known medical indication according to TTD database, and their ratio in terms of percentage (appearing in bold). (b) The bar plot shows the percentage of predicted drugs with already known medical indications grouped by metric for each disease reported in the legend. Only the metric predicting a total number of drugs greater than five for a specific disease are plotted.

Figure 3

In silico efficacy of candidate repurposable drugs for malignant breast neoplasm. (a–c) Venn diagrams of the candidate drugs predicted by using mean, median, mode metrics with respect to the standard minimum one for malignant breast neoplasm treatment. (d) Bar plot showing the percentage of metric-specific candidate drugs that have a GSEA score greater than zero.

Figure 4

In silico efficacy of candidate repurposable drugs for prostate neoplasm. (a–c) Venn diagrams of the candidate drugs predicted by using mean, median, mode metrics with respect to the standard minimum one for prostate neoplasm treatment. (d) Bar plot showing the percentage of metric-specific candidate drugs that have a GSEA score greater than zero.

Figure 5

Venn diagram of the predicted repurposable drugs for each disease (a–d) according to the different exploited metrics. Metric-specific drugs with an already known relevant medical indication according to the TTD database and discussed in the text are highlighted in red.