A multispecies foundation model for genome function prediction, gene annotation and sequence design
Abstract
Comparative genomics and synthetic biology require models that can interpret, annotate, and design DNA sequences across species, yet genomic models have not combined these capabilities beyond a few well-characterized organisms. Here we introduce Nucleotide Transformer v3 (NTv3), a long-context cross-kingdom foundation model that provides a common backbone for functional prediction, gene annotation and controllable sequence generation across phylogenetically diverse species. NTv3 is pretrained on more than 128,000 genomes using sequences up to 1 Mb and post-trained on 16,262 functional tracks across 15 animal and plant genomes. NTv3 performs comparably to AlphaGenome on matched human assays and matches or exceeds evaluated published models for functional-track prediction across 19 species spanning mammals, birds, fish, plants, fungi and protists. Ablation experiments show that both genome pre-training and expanding functional supervision from human and mouse to 15 species improve track prediction. Leveraging these learned functional representations, NTv3 outperforms leading ab initio gene predictors across mammals, plants and insects, with fewer spurious gene models and improved recovery and reduced fragmentation of long genes. Finally, the same representations can be used with discrete diffusion to design regulatory DNA sequences with specific activity levels. We demonstrate this in plant and fly promoters and enhancers, and experimentally validate the activity and promoter selectivity of the generated fly enhancers. Together, NTv3 unifies genome interpretation, gene annotation and regulatory sequence design across the animal and plant kingdoms, with direct relevance to agriculture, biodiversity research and biotechnology.
Related
- Experimental benchmarking remains crucial to the interpretation of sequence annotation data
- CDS-BART: a BART-based foundation model for mRNA sequence analysis.
- An open structure-guided RNA foundation model for generalizable prediction
- GPCR-PTM: an interpretable, topology-aware rule-based framework for residue-level post-translational modification annotation and prediction across the human GPCRome
- KAT-Net: a hyperbolic routing framework for joint lysine acylation site prediction with shared sequence–structure representations
BibTeX
@misc{boshar2026multispecies,
title = {{A multispecies foundation model for genome function prediction, gene annotation and sequence design}},
author = {Sam Boshar and Benjamin Evans and Ziqi Tang and Yanis Adel and Yohan Bornachot and Shawn Sidbon and Franziska K. Lorbeer and Armand Picard and Chandana Rajesh and Evan Seitz and David Emms and Javier Mendoza-Revilla and Ariana Hernandez and Fatimah Al-Ani and Tristan Karch and Yair Schiff and Marie Lopez and Alexandre Laterre and Karim Beguir and Peter Koo and Volodymyr Kuleshov and Alexander Stark and Bernardo P. de Almeida and Thomas Pierrot},
howpublished = {bioRxiv},
year = {2026},
month = oct,
doi = {10.64898/2025.12.22.695963},
url = {https://doi.org/10.64898/2025.12.22.695963},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - UNPB
TI - A multispecies foundation model for genome function prediction, gene annotation and sequence design
AU - Sam Boshar
AU - Benjamin Evans
AU - Ziqi Tang
AU - Yanis Adel
AU - Yohan Bornachot
AU - Shawn Sidbon
AU - Franziska K. Lorbeer
AU - Armand Picard
AU - Chandana Rajesh
AU - Evan Seitz
AU - David Emms
AU - Javier Mendoza-Revilla
AU - Ariana Hernandez
AU - Fatimah Al-Ani
AU - Tristan Karch
AU - Yair Schiff
AU - Marie Lopez
AU - Alexandre Laterre
AU - Karim Beguir
AU - Peter Koo
AU - Volodymyr Kuleshov
AU - Alexander Stark
AU - Bernardo P. de Almeida
AU - Thomas Pierrot
PY - 2026
DA - 2026/10/29
PB - bioRxiv
DO - 10.64898/2025.12.22.695963
UR - https://doi.org/10.64898/2025.12.22.695963
AB - Comparative genomics and synthetic biology require models that can interpret, annotate, and design DNA sequences across species, yet genomic models have not combined these capabilities beyond a few well-characterized organisms. Here we introduce Nucleotide Transformer v3 (NTv3), a long-context cross-kingdom foundation model that provides a common backbone for functional prediction, gene annotation and controllable sequence generation across phylogenetically diverse species. NTv3 is pretrained on more than 128,000 genomes using sequences up to 1 Mb and post-trained on 16,262 functional tracks across 15 animal and plant genomes. NTv3 performs comparably to AlphaGenome on matched human assays and matches or exceeds evaluated published models for functional-track prediction across 19 species spanning mammals, birds, fish, plants, fungi and protists. Ablation experiments show that both genome pre-training and expanding functional supervision from human and mouse to 15 species improve track prediction. Leveraging these learned functional representations, NTv3 outperforms leading ab initio gene predictors across mammals, plants and insects, with fewer spurious gene models and improved recovery and reduced fragmentation of long genes. Finally, the same representations can be used with discrete diffusion to design regulatory DNA sequences with specific activity levels. We demonstrate this in plant and fly promoters and enhancers, and experimentally validate the activity and promoter selectivity of the generated fly enhancers. Together, NTv3 unifies genome interpretation, gene annotation and regulatory sequence design across the animal and plant kingdoms, with direct relevance to agriculture, biodiversity research and biotechnology.
SN - 2692-8205
ER -