Deepvirusclassifier: a deep learning tool for classifying SARS-CoV-2 based on viral subtypes within the coronaviridae...

Area of Science:

• * Computational biology and bioinformatics.
• * Genomic surveillance and viral evolution.

Background:

• * Accurate classification of viral sequences is crucial for understanding viral evolution and controlling outbreaks.
• * Similarities among coronaviruses present challenges for precise identification.
• * Rapid identification of emerging viral strains is essential for public health.

Purpose of the Study:

• * To develop DeepVirusClassifier, a deep learning tool for accurate classification of SARS-CoV-2 and other Coronaviridae sequences.
• * To provide a robust and efficient solution for viral identification and classification.
• * To advance viral genomics research and support surveillance of novel viral strains.

Main Methods:

• * Development of a one-dimensional deep convolutional neural network (CNN) model.
• * Training and testing on the Coronaviridae family, including SARS-CoV-2.
• * Performance evaluation using F1-score, AUROC, and artificial mutation tests; comparison with BLAST algorithm.

Main Results:

• * DeepVirusClassifier achieved over 99% sensitivity for sequences with <2000 mutations.
• * Demonstrated superior accuracy and significantly reduced processing times compared to BLAST.
• * Exceptional performance metrics indicate robust learning and high reliability.

Conclusions:

• * DeepVirusClassifier offers a powerful and accurate method for classifying viral sequences within the Coronaviridae family.
• * The tool's ability to handle sequence variations aids in viral classification and genomic research.
• * Holds significant promise for rapid identification of emerging viral strains in critical surveillance efforts.