Drug-pathway association prediction: from experimental results to computational models

Area of Science:

• Pharmacology and Bioinformatics
• Computational Biology
• Drug Discovery

Background:

• Complex human diseases require effective drugs, but traditional drug development is time-consuming and expensive.
• Current drug discovery often targets a single molecule, yet drugs typically affect biological pathways.
• Pathway-based drug discovery leverages the understanding that drugs modulate entire pathways, not just single targets.

Purpose of the Study:

• To review the current landscape of computational methods for predicting drug-pathway associations.
• To highlight the importance of identifying drug-pathway links for advancing pathway-based drug discovery.
• To provide an overview of available databases, computational techniques, and evaluation strategies.

Main Methods:

• Categorization of computational methods into Bayesian sparse factor-based, matrix decomposition-based, and other machine learning approaches.
• Review of publicly accessible databases and web servers relevant to drug-pathway associations.
• Discussion of various evaluation strategies for assessing predictive model performance.

Main Results:

• Identification and classification of state-of-the-art computational models for inferring drug-pathway associations.
• Listing of key resources and databases for researchers in this field.
• Summary of established evaluation metrics for computational drug-pathway association prediction.

Conclusions:

• Computational models offer a faster and more cost-effective alternative to experimental methods for identifying drug-pathway associations.
• Further advancements in data collection and computational model development are needed.
• The review provides a foundation for future research in pathway-based drug discovery.