Development of Predictive Models for Identifying Potential S100A9 Inhibitors Based on Machine Learning Methods.

Area of Science:

• Biochemistry
• Computational Chemistry
• Drug Discovery

Background:

• S100A9 is a therapeutic target for prostate cancer, colorectal cancer, and Alzheimer's disease.
• Lack of atomic-level data on S100A9 interactions hinders rational drug design.

Purpose of the Study:

• Develop predictive models for S100A9 inhibitory effects.
• Facilitate rational drug design for S100A9 inhibitors.

Main Methods:

• Applied machine learning classifiers on 2D-molecular descriptors.
• Optimized models using feature selectors and random forest classifiers.
• Screened over 6,000,000 compounds using optimized feature sets.

Main Results:

• Generated eight robust random forest models with high predictability and cost-effectiveness.
• Reduced 2,798 features to dozens, enabling efficient large-scale screening.
• Identified 46 potential S100A9 inhibitors through consensus voting.

Conclusions:

• The developed models offer high predictive power and cost-reduction for drug discovery.
• These models provide insights for designing novel S100A9-targeting drugs.
• Accelerated identification of potential S100A9 inhibitors for therapeutic applications.