Machine Learning augmented docking studies of aminothioureas at the SARS-CoV-2-ACE2 interface

Area of Science:

• Computational chemistry
• Drug discovery
• Bioactivity prediction

Background:

• The COVID-19 pandemic highlighted the need for rapid prediction of compound bioactivity.
• Computational chemistry offers tools like docking and Quantitative Structure-Activity Relationship (QSAR) modeling for this purpose.

Purpose of the Study:

• To compare different docking protocols for predicting the bioactivity of compounds.
• To identify potential SARS-CoV-2 S-protein inhibitors using computational methods.

Main Methods:

• Utilized various docking protocols to screen compounds against the SARS-CoV-2 S-protein/ACE2 interface.
• Employed Machine Learning (ML), specifically Random Forest Regressor, for large-scale prediction.
• Trained the ML model on over 1800 compounds with known binding properties.

Main Results:

• Compared the efficacy of different docking approaches in identifying potential binders.
• Successfully predicted binding properties for a large dataset of nearly 600,000 compounds.
• Identified compounds with potential bioactivity against the SARS-CoV-2 S-protein.

Conclusions:

• Docking and ML are valuable tools for rapid virtual screening in drug discovery.
• The study provides a large set of potential SARS-CoV-2 inhibitors for further investigation.
• Highlights the importance of selecting appropriate computational methods for accurate bioactivity prediction.