Obtaining an accurate estimate of the Covid-19 mutation rate via Coronavirus sequence analysis preeminent themes using convolutional neural networks

Abstract

It is believed that Wuhan is where the SARS-CoV-2 virus first surfaced in 2019. This virus was the cause of the coronavirus epidemic (COVID-19), which was officially declared a worldwide pandemic. To identify and manage COVID-19 while the disease is progressing, a great deal of research has been done. It is difficult to distinguish between COVID-19 and other coronavirus strains due to their genetic similarities. That being said, it is imperative that we quickly determine whether an epidemic is caused by a newly discovered virus or a long-standing illness. This article discusses the DeepCoV deep-learning (DL) deep learning approach, which uses layered convolutional neural networks (CNNs) to classify viral illnesses, including SARS-CoV-2. Numerous SARS-CoV-2-related motifs may also be found by carefully examining the computational filter procedures. By spotting these important patterns, DeepCoV makes CNN transparent. The research was done using the 2019nCoVR datasets, and the findings show that DeepCoV outperformed other benchmark machine learning models in terms of accuracy. DeepCoV also achieved maximum scores of 98.58 % and 98.62 %, respectively, on the receiver operating characteristic curve (AUC-ROC) and precision-recall curve (AUCPR). Furthermore linked to the coronavirus genome are notable patterns, filter activities, and particular feature activation actions. When combined, these results provide important insights into how to apply deep learning (DL) algorithms in lieu of SARS-CoV-2 detection and illness patterns in the SARS-CoV-2 genes.