Online Predictor Using Machine Learning to Predict Novel Coronavirus and Other Pathogenic Viruses

Area of Science:

• Virology
• Epidemiology
• Bioinformatics
• Machine Learning

Background:

• Virus classification is a critical challenge in virology and epidemiology.
• Accurate identification of novel coronaviruses is essential for public health.
• Machine learning offers a promising approach for sequence-based virus prediction.

Purpose of the Study:

• To develop and validate a machine learning model for predicting novel coronavirus sequences.
• To create a tool, COVID-Predictor, for classifying SARS-CoV-1, MERS-CoV, and SARS-CoV-2.
• To assess the model's performance against other machine learning techniques.

Main Methods:

• Trained a Naive Bayes classifier using 1000 viral sequences (SARS-CoV-1, MERS-CoV, SARS-CoV-2, and others).
• Employed 10-fold cross-validation for robust model assessment.
• Developed a web-based application for uploading unknown sequences for prediction.

Main Results:

• Achieved an average accuracy of 99.7% on an unseen validation dataset.
• Demonstrated the superiority of the COVID-Predictor model compared to other machine learning methods.
• Successfully deployed the pre-trained model into a user-friendly web application.

Conclusions:

• The COVID-Predictor model provides highly accurate and efficient coronavirus prediction.
• Machine learning, specifically the Naive Bayes classifier, is effective for viral sequence analysis.
• The web application facilitates rapid identification of unknown coronavirus sequences, aiding in disease surveillance.