Deep Learning Approach for Discovery of In Silico Drugs for Combating COVID-19

Area of Science:

• Computational biology
• Artificial intelligence in medicine
• Drug discovery

Background:

• Early diagnosis and effective treatments are crucial for managing pandemic diseases like COVID-19.
• Current treatments have shown limited efficacy, necessitating novel therapeutic strategies.
• Deep learning has shown significant potential in accelerating medical research and vaccine development.

Purpose of the Study:

• To develop a deep learning approach for identifying potential drug candidates against COVID-19.
• To utilize Quantitative Structure-Activity Relationship (QSAR) modeling for drug target identification and binding affinity calculation.
• To train deep learning models on molecular descriptor datasets for robust drug discovery and feature extraction.

Main Methods:

• Application of machine learning algorithms including logistic regression, Support Vector Machines (SVM), and Random Forest.
• Quantitative Structure-Activity Relationship (QSAR) modeling for predicting protein-ligand interactions and binding affinities.
• Deep learning models trained on molecular descriptors for feature extraction and drug candidate identification.

Main Results:

• The study identified numerous molecules exhibiting significant binding affinities (greater than -18) against SARS-CoV-2.
• These molecules demonstrate potential in inhibiting the replication of the virus responsible for COVID-19.
• The deep learning approach facilitated robust drug discovery and feature extraction for combating the pandemic.

Conclusions:

• The proposed deep learning and QSAR modeling approach is effective for discovering novel drug candidates against COVID-19.
• Identified molecules show promise for blocking SARS-CoV-2 multiplication, potentially leading to new therapeutic interventions.
• This methodology can accelerate the development of treatments for current and future pandemic diseases.