Predicting Antiviral Drugs for COVID-19 Treatment Using Artificial Intelligence Based Approach

ABSTRACT

Today’s world is seriously affected and medical infrastructure is challenged due to the novel COVID-19 epidemics. On 15th May 2020, Naturemedicine published a detailed survey report which concluded that drugs like aminoquinolines, chloroquine, hydroxychloroquine targeted the virus replication cycle by keeping the virus out of host cells and helping people to overcome the disease. But, the exponential incidence rate cannot be stopped and some limitations also detected to treat the patients using these drugs. In these circumstances, suitable medicines are immediately needed that can also help in vaccine production. In this chapter, a novel AI based architecture has been proposed to obtain influencial drugs for the COVID-19 treatment. Here, The COVID-19 (MT050943 and MT012098) data are taken from the NCBI database, submitted by the Government of India to build tests and find drugs. A detailed analysis of these sequences has been performed to find the noncoding percentage, gene density, and abundances in functional gene categories using Virus Amplicon Sequencing Assembly Pipeline (V-ASAP). This analysis finds the unique regions in the virus genome as a target region. This knowledge is feed into the AI model. Firstly, the informative genes are selected from the target region. Then the Advanced Matched Molecular Pair (AMMP) analysis is performed to study the possible local region of a drug and its impact. At last, a Generative Adversarial Convolution Neural Network (GACNN) is trained with the information gain and performs a gradient-based optimization to predict the drug to target for the validate genome. The result of the proposed methodology obtains higher accuracy than the other existing methods and identified effective drugs (like, Dexamethasone, Remdesivir etc.) for the study of the COVID-19 vaccine. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.