Genos: a human-centric genomic foundation model
GigaScience · 06.01.2025 · 14
DOI: 10.1093/gigascience/giaf132 ↗Kurz gefasst
Fachartikel in der Zeitschrift GigaScience, online erschienen am 01.01.2026. Verfasst von Adi Lin und 59 weiteren Autor:innen. Beteiligte Einrichtung: Genos Team, West Lake District, Hangzhou 310012, China. Hauptthemen laut MEDLINE: Genomics, Genome, Human, Software, Computational Biology. Die Arbeit wurde bisher 19-mal zitiert (Crossref) und nennt 33 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Background
The rapid expansion of human genomic data demands foundation models that manage ultra-long sequences and capture population diversity, limitations common in existing models that lack human-specific representation, and clinical inference efficiency.
Results
Here, we introduce Genos (Genos-1.2B/Genos-10B), a human-centric genomic foundation model engineered for million-basepair sequence modeling. Genos utilizes a large-scale mixture of experts structure, optimized for a 1-Mb context, trained on high-quality human de novo assemblies from datasets such as the Human Pangenome Reference Consortium and the Human Genome Structural Variation Consortium, representing diverse global populations. A suite of optimization strategies was implemented to ensure training stability and enhance computational efficiency, which collectively reduces costs and facilitates million-basepair context modeling. Functionally, Genos performs single-nucleotide resolution analysis and dynamically simulates the cascade effects of noncoding variations on RNA expression profiles. In comprehensive evaluations, Genos uniformly surpasses state-of-the-art models on critical human genomics benchmarks and demonstrates robust omics-text cross-modal diagnostic capabilities. We present a systematic technical evaluation and validation of Genos’s architecture, training convergence, and performance across standard benchmarks.
Conclusions
This work provides a reliable technical blueprint and performance benchmark for the development of the next generation of high-efficiency genomic foundation models. Genos model weights, inference code, and usage documentation are publicly available on GitHub (https://github.com/BGI-HangzhouAI/Genos) and Hugging Face Hub (https://huggingface.co/BGI-HangzhouAI).
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Genos Team, West Lake District, Hangzhou 310012, China.
ORCID-Profile: Duoyuan Chen, Jie Hu, Ling Guo, Shiping Liu, Xin Jin, Xinjiang Tan, Xun Xu, Zhan Xiao, Zhaorong Li
Zitieren
APA
Lin, A., Xie, B., Ye, C., Wang, C., Chen, D., Wang, E., Lu, F., Xue, G., Zhang, H., Zhan, J., Zhang, J., Pang, J., Liang, J., Lin, J., Ma, J., Hu, J., Ma, J., Dong, J., Li, J., …, & Zhao, Z. (2025). Genos: a human-centric genomic foundation model. GigaScience, 14. https://doi.org/10.1093/gigascience/giaf132
Vancouver
Lin A, Xie B, Ye C, Wang C, Chen D, Wang E, et al. Genos: a human-centric genomic foundation model. GigaScience. 2025;14. doi:10.1093/gigascience/giaf132
BibTeX
@article{Lin2025,
author = {Lin, Adi and Xie, Bin and Ye, Cheng and Wang, Cheng and Chen, Duoyuan and Wang, Ercheng and Lu, Fanfeng and Xue, Guirong and Zhang, Haiqiang and Zhan, Jiajie and Zhang, Jianfeng and Pang, Jiangshuan and Liang, Jianqiang and Lin, Jiawei and Ma, Jiaxin and Hu, Jie and Ma, Jing and Dong, Jinni and Li, Jiongzhen and Liu, Junchen and Chen, Junhong and Li, Junyou and Ding, Kai and Deng, Kaiwen and Chen, Kui and Wang, Lihui and Liu, Longqi and Guo, Ling and Xiong, Liwen and Yang, Luhao and Cheng, Ming and Chen, Nanning and Chen, Renzhong and Sun, Shanxin and Li, Shaoshuai and Chen, Shicheng and Liu, Shiping and Xie, Siwei and Liu, Suyan and Zhou, Tao and Tang, Wangyang and Zhang, Weiqiang and Jiang, Xianyue and Qi, Xianzhi and Jin, Xin and Tan, Xinjiang and Hu, Xinyue and Xu, Xun and Feng, Xuyang and Lu, Yafei and Gao, Yifan and Shang, Yong and He, Youzhe and Yuan, Yue and Wang, Yufan and Liu, Yuqi and Xiao, Zhan and Meng, Zhangyuan and Li, Zhaorong and Zhao, Zhe},
title = {Genos: a human-centric genomic foundation model},
journal = {GigaScience},
year = {2025},
volume = {14},
doi = {10.1093/gigascience/giaf132}
}
EUVIMED