Craft: A Machine Learning Approach to Dengue Subtyping

Abstract

Motivation: The dengue virus poses a major global health threat, with nearly 390 million infections annually. A recently proposed hierarchical dengue nomenclature system enhances spatial resolution by defining major and minor lineages within genotypes, aiding efforts to track viral evolution. While current subtyping tools - Genome Detective, GLUE, and NextClade - rely on computationally intensive sequence alignment and phylogenetic inference, machine learning presents a promising alternative for achieving accurate and rapid classification.

Results: We present Craft (Chaos Random Forest), a machine learning framework for dengue subtyping. We demonstrate that Craft is capable of faster classification speeds while matching or surpassing the accuracy of existing tools. Craft achieves 99.5% accuracy on a hold-out test set and processes over 140 000 sequences per minute. Notably, Craft maintains remarkably high accuracy even when classifying sequence segments as short as 700 nucleotides.

Fig. 1:

Composite radar-chart of class-wise model performance. Classes are colored according to serotype and genotype. The blue, red and green dotted lines represent Craft, NextClade and Genome Detective respectively, with lines closer to the perimeter indicating better performance. The length of each bar corresponds to the evolutionary depth of the lineage.

Fig. 2:

A collection of line plots showing the accuracy of each model when tested on short genomic segments from various positions within the dengue genome. Each plot corresponds to a particular serotype. We include horizontal bars indicating the positions of each gene region for each serotype.