September 11, 2025Open Access

A novel Vector-Symbolic Architecture for graph encoding and its application to viral pangenome-based species classification

View Full Paper

Ask AI

Bookmark

Authors

Discussion

This research employs vector-symbolic architecture to enhance viral classification accuracy, indicating potential in genomic taxonomy.

The flat species-level model achieved 87.08% classification accuracy, highlighting its strength in viral classification.
Model architecture significantly impacts classification accuracy; hierarchical approaches led to 33.57% accuracy, revealing broad taxonomic challenges.
Hyperdimensional computing using weighted de Bruijn graphs allows effective encoding of multi-species viral pangenomes.
Despite its promise, alignment-free classification presents challenges due to signal dilution and error propagation in complex models.

Cumbo et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark