Drug discovery through Covid-19 genome sequencing with siamese graph convolutional neural network

Area of Science:

• Virology
• Computational Biology
• Pharmacology

Background:

• The COVID-19 pandemic caused significant global mortality due to evolving viral variants.
• Neural network-based drug discovery methods for COVID-19 face challenges like complexity and convergence issues.
• Existing methods identified ineffective drugs, highlighting the need for improved approaches.

Purpose of the Study:

• To propose a hybrid computational architecture for effective COVID-19 drug repurposing.
• To identify optimal drug candidates for COVID-19 treatment by analyzing viral genetic sequences.
• To overcome the limitations of traditional neural network approaches in drug discovery.

Main Methods:

• Investigated viral gene density and noncoding proportions using next-generation sequencing.
• Identified Drug Target Regions (DTRs) within the virus DNA sequence.
• Applied variable DNA neighborhood search to create DNA interaction networks and utilized D3Similarity for drug database analysis.

Main Results:

• The hybrid architecture successfully identified potential drug candidates.
• Remdesivir and Dexamethasone demonstrated high efficacy rates of 97.41% and 97.93%, respectively.
• The method effectively distinguished from previously proposed but ineffective treatments like hydroxychloroquine.

Conclusions:

• The proposed hybrid architecture offers a robust solution for COVID-19 drug repurposing.
• Remdesivir and Dexamethasone are computationally validated as potent therapeutic agents against COVID-19.
• This approach provides a reliable framework for identifying effective treatments for viral diseases.