AI Paper Atlas

Research map / Probabilistic and scientific modeling / Molecules and proteins

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

Arnav Shah, Junzhe Li, Parsa Idehpour, Adibvafa Fallahpour, Brandon Wang, Sukjun Hwang, BO WANG, Patrick D Hsu, Hani Goodarzi, Albert Gu

ICML 2026 · Probabilistic and scientific modeling › Molecules and proteins · approach: Biological sequence transformers · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use hierarchical dynamic chunking (dnaHNet) instead of StripedHyena2genomic autoregressive model · GTDB perplexity scalingscaling exponent alpha: 0.04 -> 0.06Improved
use dnaHNet instead of StripedHyena2 at matched computegenomic model · evaluation perplexityperplexity gap: 0.078 -> 0.118 points as compute growsImproved
use dnaHNet 2-stage (218M) instead of SH2 (166M)inference · 1M nucleotidesFLOPs: 3.89x fewer FLOPsImproved
use dnaHNet-XL instead of StripedHyena2-234Minference · 262K nucleotideslatency: 2,352 ms -> 410 msImproved
use dnaHNet instead of StripedHyena2 compute-matchedzero-shot VEP · MaveDB E. coliSpearman correlationImproved
use dnaHNet instead of StripedHyena2zero-shot gene essentiality · DEGAUROCImproved
codon-aligned (3,2) compression instead of (2,2) (ablation)dnaHNet · DEGAUROCImproved
increase compression to (3,3), 9x (ablation)dnaHNet · MaveDB VEPSpearman correlationHurt
use (2,2) instead of (3,2) compression (ablation)dnaHNet · MaveDB VEPSpearman correlationNo clear effect
go beyond two recursive stages (ablation)dnaHNet · hierarchy depthstability and returnsNo clear effect
raise compression regularization coefficient alpha to >=0.03 (ablation)dnaHNet · next nucleotide predictionfaithfulness to compression targetHurt
allocate ~30% of parameters to encoder-decoder instead of 15% (ablation)dnaHNet · genomic modelingmodel qualityImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy