Efficient Identification of Anti-SARS-CoV-2 Compounds Using Chemical Structure- and Biological Activity-Based...

Area of Science:

• Drug discovery and development
• Computational biology
• Virology

Background:

• Traditional high-throughput screening (HTS) for anti-SARS-CoV-2 compounds faces limitations in cost and efficiency.
• Identifying novel inhibitors for viral entry and 3CL protease is crucial for combating SARS-CoV-2.

Purpose of the Study:

• To develop and validate machine learning (ML) models for identifying SARS-CoV-2 inhibitors.
• To improve the hit rate and efficiency of compound screening compared to traditional HTS.

Main Methods:

• Development of ML classification models to predict compounds inhibiting SARS-CoV-2 entry or 3CL protease activity.
• Utilizing area under the receiver operating characteristic curve (AUC-ROC) to evaluate model performance.
• Experimental validation of ML-identified compounds using live SARS-CoV-2 assays.

Main Results:

• Optimal ML models demonstrated good performance with AUC-ROC values exceeding 0.78.
• ML models increased hit rates by 2.1-fold for viral entry inhibitors and 10.4-fold for 3CL protease inhibitors.
• Twenty-two compounds exhibited potent (<5 microM) antiviral activity in live virus assays.

Conclusions:

• ML models serve as a valuable complementary tool to HTS in anti-SARS-CoV-2 drug discovery.
• This approach enhances compound screening capacity, speed, and efficiency.
• ML-driven strategies can accelerate the identification of effective antiviral therapies.