A review of SARS-CoV-2 drug repurposing: databases and machine learning models

Area of Science:

• Computational biology and bioinformatics
• Drug discovery and development
• Infectious disease research

Background:

• The global threat of Severe Acute Respiratory Syndrome Corona Virus 2 (SARS-CoV-2) necessitates rapid development of effective antiviral strategies.
• Drug repurposing offers a time- and cost-efficient approach to identify new therapeutic applications for existing FDA-approved drugs.
• Machine learning (ML) has shown significant promise in accelerating virtual drug screening and identifying potential inhibitors.

Purpose of the Study:

• To review frequently utilized databases for ML-based drug repurposing studies targeting SARS-CoV-2.
• To survey recent ML models, including Deep Learning and conventional approaches, for predicting potential SARS-CoV-2 inhibitors.
• To guide researchers in selecting appropriate databases by detailing their features and limitations.

Main Methods:

• Literature review of databases commonly employed in ML-driven SARS-CoV-2 drug repurposing.
• Systematic review of recent ML models (Deep Learning and conventional) applied to predict SARS-CoV-2 inhibitors.
• Analysis of methodologies, applications, features, and limitations of reviewed databases and ML models.

Main Results:

• Identified key databases crucial for data extraction in ML-based SARS-CoV-2 drug repurposing.
• Summarized the performance and applicability of various ML models in predicting potential SARS-CoV-2 inhibitors.
• Provided insights into the strengths and weaknesses of different data resources for research.

Conclusions:

• ML, coupled with appropriate databases, is a powerful strategy for efficient drug repurposing against SARS-CoV-2.
• The reviewed databases and ML models offer valuable resources for researchers aiming to discover novel antiviral therapies.
• Informed database selection is critical for the success of ML-driven drug repurposing initiatives.