QSAR Regression Models for Predicting HMG-CoA Reductase Inhibition

Area of Science:

• Medicinal Chemistry
• Computational Chemistry
• Pharmacology

Background:

• HMG-CoA reductase is a key enzyme in cholesterol synthesis.
• Inhibitors of HMG-CoA reductase are crucial for treating cardiovascular diseases.

Purpose of the Study:

• To develop and validate quantitative structure-activity relationship (QSAR) models for human HMG-CoA reductase inhibitors.
• To identify novel potential inhibitors of HMG-CoA reductase.
• To explore the application of QSAR models in understanding herbal extract bioactivity.

Main Methods:

• Utilized nested cross-validation for QSAR model validation.
• Employed machine learning regression algorithms, feature selection, and molecular descriptors/fingerprints.
• Screened over 220,000 compounds from the ZINC 15 database using validated QSAR models.

Main Results:

• Developed 21 high-performing QSAR models (R² ≥ 0.70 or CCC ≥ 0.85).
• Constructed five ensemble models from the top six QSAR models.
• Identified 237 compounds with predicted IC50 values ≤ 10 nM, including novel potential inhibitors.
• An svm-based ensemble model predicted potent inhibitors for approximately 0.08% of screened compounds.

Conclusions:

• Developed accurate QSAR models for predicting HMG-CoA reductase inhibitors.
• The models can accelerate the discovery of new cholesterol-lowering drugs.
• QSAR models show potential for elucidating the cholesterol-lowering mechanisms of herbal extracts.