# Host centric drug repurposing for viral diseases

## TL;DR

This paper introduces a new computational method to repurpose drugs by targeting host cell factors involved in viral infections, rather than the virus itself.

## Contribution

A novel approach combining collaborative filtering and network medicine for host-centric drug repurposing against viruses.

## Key findings

- The method successfully identifies known antiviral drugs among top predictions.
- The model captures meaningful biological insights into viral infection mechanisms.
- Predictions are generated for 143 viruses using a generalizable framework.

## Abstract

Computational approaches for drug repurposing for viral diseases have mainly focused on a small number of antivirals that directly target pathogens (virus centric therapies). In this work, we combine ideas from collaborative filtering and network medicine for making predictions on a much larger set of drugs that could be repurposed for host centric therapies, that are aimed at interfering with host cell factors required by a pathogen. Our idea is to create matrices quantifying the perturbation that drugs and viruses induce on human protein interaction networks. Then, we decompose these matrices to learn embeddings of drugs, viruses, and proteins in a low dimensional space. Predictions of host-centric antivirals are obtained by taking the dot product between the corresponding drug and virus representations. Our approach is general and can be applied systematically to any compound with known targets and any virus whose host proteins are known. We show that our predictions have high accuracy and that the embeddings contain meaningful biological information that may provide insights into the underlying biology of viral infections. Our approach can integrate different types of information, does not rely on known drug-virus associations and can be applied to new viral diseases and drugs.

Drug repurposing is the re-use of de-risked compounds in humans for new therapeutic indications. Computational approaches can help in this process by providing a ranking of compounds with potential effect against a given disease. For viral diseases, computational methods have mainly focused on a small number of antivirals that directly target pathogens (virus centric therapies). Another type of antiviral drugs is aimed at disrupting host cellular processes required for the viral infection, either directly or indirectly (host centric therapies). In this work, we propose a novel computational approach focused on host centric therapies that can make predictions on a large set of drugs. It relies on the human protein-protein interaction network for identifying drugs that may influence host cellular processes required for the viral infection, together with machine learning techniques to enhance prediction accuracy by integrating information across different viruses and drugs. We obtained predictions of drug efficacy for 143 viruses. We show that our method successfully identifies drugs with known evidence against viruses among the top-ranked predictions. Furthermore, our model captures meaningful biological information, providing insights into the underlying biology of viral infections.