Real time classification of viruses in 12 dimensions

Area of Science:

• Virology
• Bioinformatics
• Computational Biology

Background:

• Current viral classification systems face challenges with completeness and accuracy, as evidenced by significant missing label rates in databases like GenBank.
• Existing methods for determining viral phylogenetic relationships are often complex and may suffer from issues of uniqueness and existence in visualization.

Purpose of the Study:

• To develop a novel computational approach for the taxonomic classification and phylogenetic analysis of viruses.
• To address the limitations of current viral classification methods by providing a robust and real-time analytical framework.

Main Methods:

• Development and application of a Natural Vector representation to embed 2,044 viral genomes from GenBank into a computable space.
• Utilizing a unique graphical representation for virus phylogeny, distinct from traditional tree-based methods, for visualization and analysis.

Main Results:

• The Natural Vector approach enables real-time determination of phylogenetic relations across all viral taxonomic levels (Baltimore class to species).
• Achieved high prediction accuracy rates: 98.2% for Baltimore class, 96.6% for family, 99.7% for subfamily, and 97.2% for genus.
• Successfully predicted and corrected viral classification information and identified viral origins, such as the proximity of West Nile virus to the Japanese encephalitis antigenic complex.

Conclusions:

• The proposed Natural Vector representation and graphical method offer a unique, stable, and accurate solution for viral taxonomy and phylogeny.
• This approach significantly enhances the ability to classify viruses, understand their evolutionary relationships, and identify potential public health threats.