DeepD enables efficient genome modeling and controllable long-context sequence generation across domains of life
- 1China Agricultural University
Abstract
Genome foundation models hold great potential for advancing biological research, but modeling diverse complex genomes and generating controllable genome-scale sequences remain challenging. We present DeepD, a phylogeny-informed sparse mixture-of-experts model operating at single-nucleotide resolution over a 1-million-token context while reducing active computation. Pretrained on unselected whole-genome sequences across all domains of life, DeepD learns genic and intergenic sequence features and supports broad sequence-to-function prediction across species. DeepD performs well in human protein fitness and disease variants effect prediction, such as BRCA1 and BRCA2, with particular strength in regulatory regions, such as promoter and transcription-factor binding sites. Expert routing captures exon-intron boundaries, transposable elements, and distal cis-regulatory elements. DeepD also generates experimentally validated core promoters with specified transcriptional strengths. Finally, the Context-Anchor-Body strategy enables generation of complete yeast chromosome sequences and megabase-scale Arabidopsis segments, preserving gene structure and order while introducing variations within natural ranges. Together, DeepD is an efficient genome foundation model that unifies accurate sequence-to-function prediction with controllable genome-scale generation, advancing predictive and generative design of complex genomes.
Related
- Controlled activin A signaling enables liver organoid generation from human pluripotent stem cells
- seqproc: An efficient, flexible, and concise tool for sequence geometry description and transformation
- Targeting IL-2 to inflamed tissues via oxidation-specific epitopes enables third-generation bispecific IL-2 therapeutics
- Efficient primer design for broad-range virus detection: A semi-automated workflow using sequence clustering and varVAMP
- DeepSCENIC: transfer learning from sequence-to-function models enables causal gene regulatory network inference
BibTeX
@misc{lai2026deepd,
title = {{DeepD enables efficient genome modeling and controllable long-context sequence generation across domains of life}},
author = {J. Lai and B. Xin and Z. Wang and A. Yu and Z. Chen and Y. Guo and M. Zhang and B. Hu and C. Ye and J. Chen and Q. Liu and H. Zhao and W. Song and N. Dong and F. Yang and Q. Yang and H. Zhao},
howpublished = {bioRxiv},
year = {2026},
month = oct,
doi = {10.64898/2026.10.13.758170},
url = {https://www.biorxiv.org/content/10.64898/2026.10.13.758170v1},
note = {CC BY-NC},
}
RIS
TY - UNPB
TI - DeepD enables efficient genome modeling and controllable long-context sequence generation across domains of life
AU - J. Lai
AU - B. Xin
AU - Z. Wang
AU - A. Yu
AU - Z. Chen
AU - Y. Guo
AU - M. Zhang
AU - B. Hu
AU - C. Ye
AU - J. Chen
AU - Q. Liu
AU - H. Zhao
AU - W. Song
AU - N. Dong
AU - F. Yang
AU - Q. Yang
AU - H. Zhao
PY - 2026
DA - 2026/10/13
PB - bioRxiv
DO - 10.64898/2026.10.13.758170
UR - https://www.biorxiv.org/content/10.64898/2026.10.13.758170v1
AB - Genome foundation models hold great potential for advancing biological research, but modeling diverse complex genomes and generating controllable genome-scale sequences remain challenging. We present DeepD, a phylogeny-informed sparse mixture-of-experts model operating at single-nucleotide resolution over a 1-million-token context while reducing active computation. Pretrained on unselected whole-genome sequences across all domains of life, DeepD learns genic and intergenic sequence features and supports broad sequence-to-function prediction across species. DeepD performs well in human protein fitness and disease variants effect prediction, such as BRCA1 and BRCA2, with particular strength in regulatory regions, such as promoter and transcription-factor binding sites. Expert routing captures exon-intron boundaries, transposable elements, and distal cis-regulatory elements. DeepD also generates experimentally validated core promoters with specified transcriptional strengths. Finally, the Context-Anchor-Body strategy enables generation of complete yeast chromosome sequences and megabase-scale Arabidopsis segments, preserving gene structure and order while introducing variations within natural ranges. Together, DeepD is an efficient genome foundation model that unifies accurate sequence-to-function prediction with controllable genome-scale generation, advancing predictive and generative design of complex genomes.
ER -