Classification of group A rotavirus VP7 and VP4 genotypes using random forest

Area of Science:

• Virology
• Bioinformatics
• Machine Learning

Background:

• Group A rotaviruses cause severe diarrhea in children and animals globally.
• Increasing availability of rotavirus sequence data necessitates efficient genotyping methods.
• Machine learning approaches have not been extensively explored for rotavirus genotyping.

Purpose of the Study:

• To explore the efficacy of machine learning, specifically random forest models, for classifying group A rotavirus genotypes.
• To develop an efficient and accurate method for genotyping rotaviruses using sequence data.

Main Methods:

• Trained random forest models on positional features from pairwise and multiple sequence alignments.
• Employed repeated 10-fold cross-validation and leave-one-out cross-validation for model assessment.
• Validated models on unseen test datasets to evaluate real-world performance.

Main Results:

• Random forest models demonstrated high accuracy (0.972-0.996) and kappa values (0.969-0.996) in classifying VP7 and VP4 genotypes.
• Multiple sequence alignment models showed slightly higher accuracy but were slower computationally than pairwise alignment models.
• Repeated 10-fold cross-validation was faster than leave-one-out cross-validation with comparable accuracy.

Conclusions:

• Random forest models are highly effective for classifying group A rotavirus VP7 and VP4 genotypes.
• These models provide a rapid and accurate solution for analyzing the growing volume of rotavirus sequence data.
• The proposed machine learning approach enhances rotavirus genotyping capabilities.