Taxonova: Deep learning enables fast and confidence-aware microbiome species-level amplicon classification
4 affiliations
- 1School of Microbiology, University College Cork, Cork, Ireland
- 2APC Microbiome Ireland, University College Cork, Cork, Ireland
- 3Seqbiome Ltd., Cork, Ireland
- 4Department of Electrical and Electronic Engineering, University College Cork, Cork, Ireland
Abstract
Accurate species-level taxonomic classification of microbial amplicon sequences remains challenging due to the difficulty of resolving subtle variations within conserved ribosomal regions. Traditional similarity-based methods struggle with short amplicons and sequences with poor database representation. No existing method performs robustly across the full spectrum of amplicon lengths, limiting applicability across diverse sequencing platforms and variable-region targets. We present Taxonova, a deep learning framework for both short- and long-read ribosomal amplicons. Evaluated across physical and biome-simulated mock communities representing seven biomes (via the accompanying benchmark tool Taxobench), Taxonova attains species-level F-measures of 0.76–1.00 (median 0.94) for short-read and 0.82–0.99 (median 0.97) for long-read amplicons spanning 250 bp to 4,900 bp, reduces the species-level abundance-estimation error (L1 norm) by a median of 0.52 to 1.50 depending on amplicon region, and has a species-level false-positive rate that is a median of 16.2 × 10−4 lower than that of the other classifiers. Species-level assignment is expected to be most reliable for well-characterised communities, such as human-associated microbiomes. To demonstrate real-world applicability, we also compared amplicon-derived taxonomic profiles to paired shotgun whole-genome sequencing from gut, soil and vaginal microbiomes; Taxonova achieves the strongest species-level concordance (Jaccard) in all three studies, with a species-level Jaccard index 0.009 to 0.022 higher than that of the next-best classifier across both high- and low-complexity communities. Taxonova runs at a throughput of roughly 100 sequences per second on a single CPU core, making it suitable for routine deployment in research settings. Author summary: We developed Taxonova to address a long-standing problem in microbiome research: extracting species-level information from short DNA marker sequences. Taxonova abstains when the amplicon region carries too little signal to separate close relatives, returning a genus-level label rather than a guessed species. Taxonova made fewer confident mistakes when tested against seven existing tools on physical mock communities, biome-simulated mocks, and paired amplicon–shotgun datasets, Taxonova runs on ordinary laptops without specialised hardware and is freely available as both a Python package and a web application, lowering the barrier for researchers to obtain species-rank labels from their own amplicon data.
Related
- Taxonomic classification cost tracks neither sequencing depth nor community richness at single-sample scale: a measured resource protocol for 16S rRNA amplicon pipelines
- Constructing microbiome co-occurrence networks with confidence: A conditional, nonparametric, inference-based approach
- Decision Confidence Neuron in Echo State Network for Continual Evaluation of EEG Motor Imagery Classification Quality
- Enhancement of gut-microbiome-based classification of prodromal Parkinson's Disease cases by hyperparameter tuning, integration of multimodal data, and meta-analysis.
- Corpusome, a cross-body-site human microbiome corpus for representation learning
BibTeX
@article{kakabra2026taxonova,
title = {{Taxonova: Deep learning enables fast and confidence-aware microbiome species-level amplicon classification}},
author = {Kardokh Kakabra and Shriram Patel and Andriy Temko and Marcus J. Claesson},
journal = {PLOS Computational Biology},
year = {2026},
month = oct,
doi = {10.1371/journal.pcbi.1014912},
url = {https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1014912},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Taxonova: Deep learning enables fast and confidence-aware microbiome species-level amplicon classification
AU - Kardokh Kakabra
AU - Shriram Patel
AU - Andriy Temko
AU - Marcus J. Claesson
PY - 2026
DA - 2026/10/29
JO - PLOS Computational Biology
DO - 10.1371/journal.pcbi.1014912
UR - https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1014912
AB - Accurate species-level taxonomic classification of microbial amplicon sequences remains challenging due to the difficulty of resolving subtle variations within conserved ribosomal regions. Traditional similarity-based methods struggle with short amplicons and sequences with poor database representation. No existing method performs robustly across the full spectrum of amplicon lengths, limiting applicability across diverse sequencing platforms and variable-region targets. We present Taxonova, a deep learning framework for both short- and long-read ribosomal amplicons. Evaluated across physical and biome-simulated mock communities representing seven biomes (via the accompanying benchmark tool Taxobench), Taxonova attains species-level F-measures of 0.76–1.00 (median 0.94) for short-read and 0.82–0.99 (median 0.97) for long-read amplicons spanning 250 bp to 4,900 bp, reduces the species-level abundance-estimation error (L1 norm) by a median of 0.52 to 1.50 depending on amplicon region, and has a species-level false-positive rate that is a median of 16.2 × 10−4 lower than that of the other classifiers. Species-level assignment is expected to be most reliable for well-characterised communities, such as human-associated microbiomes. To demonstrate real-world applicability, we also compared amplicon-derived taxonomic profiles to paired shotgun whole-genome sequencing from gut, soil and vaginal microbiomes; Taxonova achieves the strongest species-level concordance (Jaccard) in all three studies, with a species-level Jaccard index 0.009 to 0.022 higher than that of the next-best classifier across both high- and low-complexity communities. Taxonova runs at a throughput of roughly 100 sequences per second on a single CPU core, making it suitable for routine deployment in research settings. Author summary: We developed Taxonova to address a long-standing problem in microbiome research: extracting species-level information from short DNA marker sequences. Taxonova abstains when the amplicon region carries too little signal to separate close relatives, returning a genus-level label rather than a guessed species. Taxonova made fewer confident mistakes when tested against seven existing tools on physical mock communities, biome-simulated mocks, and paired amplicon–shotgun datasets, Taxonova runs on ordinary laptops without specialised hardware and is freely available as both a Python package and a web application, lowering the barrier for researchers to obtain species-rank labels from their own amplicon data.
SN - 1553-7358
ER -