Validation guidelines for drug-target prediction methods.

Area of Science:

• Computational chemistry
• Pharmacology
• Bioinformatics

Background:

• Drug-target interactions are key for drug discovery and repurposing, informing mechanisms of action and optimizing drug profiles.
• Computational methods are widely used to predict these interactions, but validation methods vary significantly.
• Understanding the evolution of these methods and their validation is crucial for advancing drug discovery.

Purpose of the Study:

• To analyze the trends in computational drug-target interaction prediction methods over the past decade.
• To investigate the evolution of validation strategies and performance metrics in this field.
• To identify prevalent experimental validation protocols and recommend future directions.

Main Methods:

• A systematic literature review of 3,286 articles on drug-target interaction prediction published in the last 10 years.
• Natural language processing for automated abstract classification and trend analysis.
• Manual analysis of 259 studies that included experimental validation of computational predictions.

Main Results:

• A significant increase in publications on drug-target interaction prediction since 2014.
• Docking and regression are the most frequently utilized computational techniques.
• Cross-validation is a common computational validation strategy, but experimental validation is infrequent.
• Multiple, orthogonal validation strategies are recommended for computational predictions.

Conclusions:

• The field of computational drug-target interaction prediction has grown substantially, with established computational methods and validation approaches.
• There is a critical need for more routine and rigorous experimental validation to confirm the biological relevance of computational predictions.
• Standardized reporting of diverse validation strategies is essential for reliable drug discovery and repurposing efforts.