Related Experiment Video
Updated: Aug 5, 2026
14:34
A Bilingual Computational Workflow for Identifying Potential PLK1 Inhibitors in American Sign Language and English
Published on: April 3, 2026
An ML-Based QSAR Web Server for KEAP1 Inhibitor Bioactivity Prediction: Composite-Score-Driven Training and Advanced
1Department of Chemistry and Biochemistry, Louise Dilworth Davis College of Science & Engineering, Texas Christian University, Fort Worth, Texas 76129, United States.
ACS Omega
|August 1, 2026
Summary
Researchers developed a machine learning model to predict KEAP1 inhibitors, crucial for treating oxidative stress-related diseases like cardiovascular and neurodegenerative conditions. This tool aids in identifying potential drug candidates before synthesis, accelerating drug development.
Area of Science:
- Computational Chemistry
- Drug Discovery
- Machine Learning
Background:
- Cardiovascular and neurodegenerative diseases are leading causes of death, with oxidative stress as a key factor.
- The KEAP1:NRF2 pathway regulates the body's natural antioxidant response.
- Activating this pathway is a potential therapeutic strategy, but requires identifying specific KEAP1 inhibitors.
Purpose of the Study:
- To develop a machine learning model for predicting novel KEAP1 inhibitors.
- To accelerate the identification of drug candidates for oxidative stress-related diseases.
- To reduce failure risk in drug development by enabling pre-synthesis evaluation.
Main Methods:
- Generated molecular fingerprints for KEAP1 inhibitors using PaDEL, Mordred, and RDKit.
- Employed a composite-score-based feature selection method.
- Trained and rigorously validated 30 machine learning models, including CatBoost, using metrics like R², CCC, and external validation.
Main Results:
- The CatBoost model achieved high predictive power with R² values of 0.8373 (test) and 0.9548 (training).
- The model demonstrated significant improvements in concordance correlation coefficient, external validation, cross-validation, and y-scrambling tests.
- The predictive model was successfully deployed as a freely accessible web server for researchers.
Conclusions:
- The developed quantitative structure-activity relationship (QSAR) machine learning model effectively predicts KEAP1 inhibitors.
- This tool can significantly expedite the discovery of novel therapeutics for diseases linked to oxidative stress.
- The web server provides a valuable resource for researchers in drug discovery.