dnaHNet: A Scalable and Hierarchical Foundation Model for Genomic Sequence Learning

Genomic foundation models have the potential to decode DNA syntax, yet face a fundamental tradeoff. Standard subword tokenizers fragment biologically meaningful motifs such as codons and regulatory elements, while nucleotide-level models preserve biological coherence but incur prohibitive computational costs for long contexts. We introduce dnaHNet, a state-of-the-art tokenizer-free autoregressive model that segments and models genomic sequences end to end. Using a differentiable dynamic chunking mechanism, dnaHNet compresses raw nucleotides into latent tokens adaptively, balancing compression with predictive accuracy. Pretrained on prokaryotic genomes, dnaHNet outperforms leading architectures including StripedHyena2 in scaling and efficiency. This recursive chunking yields quadratic FLOP reductions, enabling $>3 \times$ inference speedup over Transformers. On zero-shot tasks, dnaHNet achieves superior performance in predicting protein variant fitness and gene essentiality, while automatically discovering hierarchical biological structures without supervision. These results establish dnaHNet as a scalable, interpretable framework for next-generation genomic modeling.

DNA contains instructions that help determine how living organisms grow, function, and respond to changes. Recent AI models can learn patterns in DNA, but they face a difficult tradeoff: reading DNA one letter at a time preserves biological meaning but is slow and expensive, while chopping DNA into fixed “words” can break apart meaningful biological units. We developed dnaHNet, an AI model that learns how to group DNA letters on its own instead of relying on a fixed vocabulary. As it reads raw DNA sequences, dnaHNet dynamically compresses nearby letters into larger learned pieces, allowing it to keep important biological structure while making long DNA sequences easier to process. In experiments on microbial genomes, dnaHNet was faster and more efficient than strong existing models, while also performing well on tasks such as predicting the effects of protein mutations and identifying genes that are essential for survival. Because the model also learns meaningful biological groupings without being explicitly told what to look for, dnaHNet offers a scalable and more interpretable path toward AI systems that can help researchers understand genomes.