KITSUNE

KITSUNE identifies the empirically optimal k-mer length for alignment-free phylogenomic analysis to improve resolution in genome-based phylogenetic inference.

Key Features:

• Systematic three-step approach: Evaluates k-mer lengths from whole-genome sequences using a structured procedure.
• Cumulative Relative Entropy (CRE): Assesses the distribution of k-mers across genomes to quantify uniqueness and informativeness.
• Average Number of Common Features (ACF): Evaluates the average number of shared k-mer features between genome pairs to identify informative k lengths.
• Observed Common Features (OCF): Measures actual observed commonalities among genomes at different k-mer lengths to ensure discriminative power.
• Metric integration: Combines CRE, ACF, and OCF to select k-mer lengths that optimize alignment-free phylogenomic analyses.
• Works on whole-genome sequences and assemblies: Applicable to whole genomes and de novo assembled bacterial genomes derived from error-prone long-read sequences and a published yeast genome.
• Empirical k values and species-specific k-mers: Identifies shortest species-specific k-mers for virus classification and reports empirical optimal k values for viruses (11), bacteria (17), and fungi (~34).

Scientific Applications:

• Phylogenetic Tree Inference: Provides k-mer lengths that improve accuracy of alignment-free phylogenetic tree construction.
• Species Identification: Enables species identification from de novo assembled bacterial genomes derived from error-prone long-read sequences and from a published yeast genome.
• Virus Classification: Identifies shortest species-specific k-mers that enable accurate virus classification.

Methodology:

KITSUNE computes Cumulative Relative Entropy (CRE), Average Number of Common Features (ACF), and Observed Common Features (OCF) from whole-genome sequences and integrates these metrics to select optimal k-mer lengths.