Finding Druggable Sites in Proteins Using TACTICS.

Area of Science:

• Computational biology
• Structural biology
• Drug discovery

Background:

• Structure-based drug discovery relies on identifying drug-binding sites on target proteins.
• Locating all potential druggable sites, especially transient or 'cryptic' ones, remains a significant challenge.

Purpose of the Study:

• To introduce TACTICS (trajectory-based analysis of conformations to identify cryptic sites), a novel machine-learning algorithm for predicting druggable protein sites.
• To demonstrate TACTICS's capability in identifying both known and previously unobserved binding pockets.

Main Methods:

• TACTICS utilizes molecular dynamics simulation data, employing k-means clustering to capture conformational heterogeneity.
• A random forest model analyzes protein motion and geometry to identify potential binding residues.
• Fragment docking scores residues within predicted binding pockets.

Main Results:

• TACTICS successfully recapitulated known small-molecule binding sites in SARS-CoV-2 protease and methyltransferase, and *Yersinia pestis* aryl carrier protein.
• The algorithm predicted novel potential binding sites not evident in existing experimental structures.
• The TACTICS code is publicly available for broader research application.

Conclusions:

• TACTICS offers a robust computational approach for identifying druggable protein sites from molecular dynamics data.
• This method enhances the discovery of novel binding pockets, crucial for advancing drug development.
• TACTICS provides a valuable tool for structure-based drug discovery and medicinal chemistry research.