Machine Learning-Driven Classification of Urease Inhibitors Leveraging Physicochemical Properties as Effective Filter...

Area of Science:

• Biochemistry and Cheminformatics
• Computational Biology and Drug Discovery

Background:

• Urease is a key enzyme in nitrogen metabolism, crucial for microorganisms like Helicobacter pylori.
• Urease inhibitors offer therapeutic potential against infections and diseases such as gastric cancer and chronic kidney disease.
• Traditional methods for identifying urease inhibitors face challenges due to emerging resistance.

Purpose of the Study:

• To develop and evaluate machine learning (ML) models for predicting urease inhibitors.
• To leverage physicochemical properties for accurate prediction of molecular bioactivity.
• To identify critical features and optimal ML strategies for classifying urease inhibitors.

Main Methods:

• Constructed a dataset of urease inhibitors from literature.
• Characterized inhibitors using physicochemical properties and performed exploratory data analysis.
• Trained 252 classification models using seven ML algorithms, three attribute selection methods, and six categorization strategies.

Main Results:

• Identified key features that effectively distinguish urease inhibitors from non-inhibitors.
• Tree-based ML algorithms (Random Forest, Decision Tree, XGBoost) showed superior performance.
• Incorporating 'chemical family type' and gray-zone categorization improved model accuracy and precision.

Conclusions:

• Machine learning models demonstrate significant potential for predicting urease inhibitors.
• The study provides a robust methodology for developing predictive biochemical models.
• Findings offer actionable insights for accelerating the discovery of novel urease inhibitors.