KITSUNE: A Tool for Identifying Empirically Optimal K-mer Length for Alignment-Free Phylogenomic Analysis

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Genomic DNA serves as a unique identifier for organisms.
• Alignment-free phylogenomic analysis uses k-mer distribution for genome comparison.
• Current k-mer length selection is often arbitrary, impacting phylogenetic accuracy.

Purpose of the Study:

• To develop a systematic approach for identifying the optimal k-mer length in phylogenomic analysis.
• To introduce KITSUNE, a tool for empirical k-mer length selection.
• To enhance the resolution and reliability of phylogenetic inferences.

Main Methods:

• KITSUNE employs a three-step approach: cumulative relative entropy (CRE), average common features (ACF), and observed common features (OCF).
• The tool assesses k-mer length optimality across diverse genome datasets.
• Validation included species identification using de novo assembled genomes and published datasets.

Main Results:

• KITSUNE identified optimal k-mer lengths of 11, 17, and ~34 for viruses, bacteria, and fungi, respectively.
• The tool demonstrated reliable species identification for bacterial and yeast genomes.
• KITSUNE accurately identified the shortest species-specific k-mers for viruses.

Conclusions:

• KITSUNE provides a systematic and empirically-driven method for optimal k-mer length selection in phylogenomics.
• The tool enhances the accuracy of phylogenetic tree inference and species identification.
• KITSUNE is a valuable resource for diverse genomic analyses across different taxa.