KmerSignificance score: A discriminative and biologically informed framework for viral k-mer prioritization
6 affiliations
- 1Department of Computer Science, Université du Québec à Montréal, Montréal, Québec, Canada
- 2Centre de recherche Azrieli du CHU Sainte-Justine, Montréal, Québec, Canada
- 3Cité des Sciences et de l’Innovation de Guinée, Conakry, Guinea
- 4Department of Microbiology, Infectiology and Immunology, Faculty of Medicine, Université de Montréal, Montréal, Québec, Canada
- 5Department of Pediatrics, Faculty of Medicine, Université de Montréal, Montréal, Québec, Canada
- 6Pathology Advanced Translational Research Unit (PATRU), Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, Georgia, United States of America
Abstract
Distinguishing closely related viral strains requires identifying genomic regions where subtle sequence differences carry biological significance. While k-mer-based approaches offer computational efficiency for genome analysis, existing methods lack standardized frameworks for evaluating which k-mers are most informative. Current selection strategies focus primarily on statistical discriminative power without integrating biological context. We introduce KmerSignificance Score (KSS), a k-mer prioritization framework combining three components: an information-theoretic method measuring strain-distinguishing capacity, an optimized amino acid substitution matrix (MIYATA_EVO) for mutation impact assessment, and a protein characterization score derived from annotations and database cross-references retrieved through UniProt. All three component scores and the integrated KSS are bounded in [0,1] by construction, providing a common numerical scale for ranking positions within a dataset. The discriminative component achieved a mean F1 of 0.881, not significantly different from chi-squared (0.876), the strongest of eight baselines. Top-ranked selections from the discriminative component nonetheless differed from those produced by chi-squared, especially for SARS-CoV-2 and HIV-1. The MIYATA_EVO matrix, optimized using a genetic algorithm, increased the composite agreement score across eight amino acid characteristics by 28.4% over the original MIYATA matrix. Across 17,471 viral proteins, the protein characterization score tracked UniProt’s five-level annotation score closely (Kendall’s τ=0.775), as expected for measures drawing on overlapping annotation content. Within each level, the continuous score provided finer numerical resolution among proteins. Literature cross-referencing on SARS-CoV-2 (278,738 sequences, 19 Pango-derived lineage labels), HIV-1 (12,223 sequences, 15 subtype labels), and human cytomegalovirus (HCMV; 399–646 sequences, 4–8 locus-specific genotypes) found that the top-ranked k-mers reported here map to established variant-defining mutations, subtype-specific polymorphisms, and genotype markers. KSS provides a standardized framework for viral k-mer prioritization with potential applications in variant surveillance, molecular epidemiology, and functional annotation. The tool is available at https://github.com/bioinfoUQAM/KmerSignificanceScore. Author summary: Identifying genetic differences between closely related viral strains is essential for pandemic preparedness, vaccine development, and understanding disease outbreaks. With millions of viral genomes now sequenced, researchers need tools that can rapidly prioritize genomic differences for biological investigation, not just identify those that are statistically distinctive. Current methods based on short genomic sequences distinguish viral strains but generally do not incorporate biological information about the encoded proteins. We developed KmerSignificance Score (KSS), a framework that ranks genomic positions by combining three types of evidence: how well they distinguish viral strains, how much the encoded amino acid changes affect protein properties, and how thoroughly the affected protein has already been studied. We placed the resulting scores on a 0-to-1 scale, so that each position can be scored on its own rather than relative to the rest of the dataset. We applied our framework to three major human pathogens (SARS-CoV-2, HIV-1, and human cytomegalovirus) and found that the top-scoring positions reported here correspond to sites with documented roles in immune evasion, drug resistance, viral fitness, and strain classification. KSS can help prioritize genomic features for surveillance of emerging variants, guide experimental validation, and support molecular epidemiology.
Related
- A likelihood-based framework for simultaneously learning both noise and growth dynamics using biologically-informed neural networks
- Biologically Informed Multi-Omics Integration Reveals Clinically Meaningful Patient Representations in Acute Myeloid Leukemia
- EG-BINN: Parameter-Efficient Ensembling of Gated Biologically Informed Neural Networks for Epigenetic Age Estimation
- BLOOM: Biologically Informed Flow Matching for Transcriptomic Virtual Embryo Modeling
- Reliable mechanistic operator recovery with biologically-informed neural networks: principles for architecture and optimisation design
BibTeX
@article{lebatteux2026kmersignificance,
title = {{KmerSignificance score: A discriminative and biologically informed framework for viral k-mer prioritization}},
author = {Dylan Lebatteux and Florian Corso and Hugo Soudeyns and Isabelle Boucoiran and Soren Gantt and Abdoulaye Baniré Diallo},
journal = {PLOS Computational Biology},
year = {2026},
month = oct,
doi = {10.1371/journal.pcbi.1014332},
url = {https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1014332},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - KmerSignificance score: A discriminative and biologically informed framework for viral k-mer prioritization
AU - Dylan Lebatteux
AU - Florian Corso
AU - Hugo Soudeyns
AU - Isabelle Boucoiran
AU - Soren Gantt
AU - Abdoulaye Baniré Diallo
PY - 2026
DA - 2026/10/27
JO - PLOS Computational Biology
DO - 10.1371/journal.pcbi.1014332
UR - https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1014332
AB - Distinguishing closely related viral strains requires identifying genomic regions where subtle sequence differences carry biological significance. While k-mer-based approaches offer computational efficiency for genome analysis, existing methods lack standardized frameworks for evaluating which k-mers are most informative. Current selection strategies focus primarily on statistical discriminative power without integrating biological context. We introduce KmerSignificance Score (KSS), a k-mer prioritization framework combining three components: an information-theoretic method measuring strain-distinguishing capacity, an optimized amino acid substitution matrix (MIYATA_EVO) for mutation impact assessment, and a protein characterization score derived from annotations and database cross-references retrieved through UniProt. All three component scores and the integrated KSS are bounded in [0,1] by construction, providing a common numerical scale for ranking positions within a dataset. The discriminative component achieved a mean F1 of 0.881, not significantly different from chi-squared (0.876), the strongest of eight baselines. Top-ranked selections from the discriminative component nonetheless differed from those produced by chi-squared, especially for SARS-CoV-2 and HIV-1. The MIYATA_EVO matrix, optimized using a genetic algorithm, increased the composite agreement score across eight amino acid characteristics by 28.4% over the original MIYATA matrix. Across 17,471 viral proteins, the protein characterization score tracked UniProt’s five-level annotation score closely (Kendall’s τ=0.775), as expected for measures drawing on overlapping annotation content. Within each level, the continuous score provided finer numerical resolution among proteins. Literature cross-referencing on SARS-CoV-2 (278,738 sequences, 19 Pango-derived lineage labels), HIV-1 (12,223 sequences, 15 subtype labels), and human cytomegalovirus (HCMV; 399–646 sequences, 4–8 locus-specific genotypes) found that the top-ranked k-mers reported here map to established variant-defining mutations, subtype-specific polymorphisms, and genotype markers. KSS provides a standardized framework for viral k-mer prioritization with potential applications in variant surveillance, molecular epidemiology, and functional annotation. The tool is available at https://github.com/bioinfoUQAM/KmerSignificanceScore. Author summary: Identifying genetic differences between closely related viral strains is essential for pandemic preparedness, vaccine development, and understanding disease outbreaks. With millions of viral genomes now sequenced, researchers need tools that can rapidly prioritize genomic differences for biological investigation, not just identify those that are statistically distinctive. Current methods based on short genomic sequences distinguish viral strains but generally do not incorporate biological information about the encoded proteins. We developed KmerSignificance Score (KSS), a framework that ranks genomic positions by combining three types of evidence: how well they distinguish viral strains, how much the encoded amino acid changes affect protein properties, and how thoroughly the affected protein has already been studied. We placed the resulting scores on a 0-to-1 scale, so that each position can be scored on its own rather than relative to the rest of the dataset. We applied our framework to three major human pathogens (SARS-CoV-2, HIV-1, and human cytomegalovirus) and found that the top-scoring positions reported here correspond to sites with documented roles in immune evasion, drug resistance, viral fitness, and strain classification. KSS can help prioritize genomic features for surveillance of emerging variants, guide experimental validation, and support molecular epidemiology.
SN - 1553-7358
ER -