Host centric drug repurposing for viral diseases.

Area of Science:

• Computational biology
• Network medicine
• Pharmacology

Background:

• Traditional antiviral drug discovery often focuses on targeting viruses directly (virus-centric therapies).
• There is a need for alternative strategies like host-centric therapies that target host factors essential for viral replication.
• Existing computational methods for drug repurposing are limited in scope and scale.

Purpose of the Study:

• To develop a computational framework for predicting repurposed drugs targeting host factors for viral diseases.
• To expand drug repurposing beyond virus-centric therapies to host-centric approaches.
• To leverage network medicine and collaborative filtering for large-scale drug-virus interaction prediction.

Main Methods:

• Constructed matrices quantifying drug and viral perturbations on human protein interaction networks.
• Employed matrix decomposition to learn low-dimensional embeddings for drugs, viruses, and proteins.
• Predicted host-centric antivirals by calculating the dot product of drug and virus embeddings.
• Validated predictions for accuracy and biological relevance.

Main Results:

• Achieved high accuracy in predicting host-centric antiviral drugs.
• Demonstrated that learned embeddings capture meaningful biological information relevant to viral infections.
• The approach successfully integrates diverse data types and does not require prior drug-virus associations.
• The method is generalizable to various compounds and viruses with known host protein interactions.

Conclusions:

• The developed computational approach offers a powerful and scalable method for identifying host-centric antiviral drugs.
• This network-based strategy provides novel insights into viral infection mechanisms and drug-host interactions.
• The framework facilitates systematic drug repurposing for emerging and existing viral diseases.