Deep Learning Based Drug Screening for Novel Coronavirus 2019-nCov

Area of Science:

• Computational biology and bioinformatics
• Drug discovery and development
• Virology and infectious diseases

Background:

• The novel coronavirus (2019-nCoV) is a global health concern, necessitating rapid development of effective antiviral drugs.
• Traditional drug development is time-consuming, making alternative methods crucial for emerging viral threats like 2019-nCoV.
• 2019-nCoV shares homology with SARS-CoV, suggesting similar therapeutic targets may be effective.

Purpose of the Study:

• To leverage deep learning for accelerated drug discovery against 2019-nCoV.
• To identify potential therapeutic agents targeting the 2019-nCoV 3C-like protease.
• To provide a list of candidate molecules for experimental validation.

Main Methods:

• Collected and analyzed 2019-nCoV RNA and protein sequences from patient data.
• Constructed a 3D protein model of the 3C-like protease using homology modeling.
• Employed a deep learning-based pipeline for large-scale virtual screening of chemical and peptide compound databases.

Main Results:

• Identified 3C-like protease as a significant therapeutic target for 2019-nCoV.
• Screened multiple compound libraries, including chemical and tripeptide databases.
• Generated a list of potential drug candidates, including Meglumine, Vidarabine, Adenosine, Ganciclovir, and a specific isoleucine-lysine-proline peptide combination.

Conclusions:

• Deep learning offers a rapid and effective approach for identifying antiviral drug candidates.
• The identified chemical ligands and peptide drugs show promise for combating 2019-nCoV.
• This study provides valuable guidance for experimental validation to accelerate the development of treatments for 2019-nCoV.