Deep contrastive learning enables genome-wide virtual screening
Science · 08.01.2026 · 391 (6781)
DOI: 10.1126/science.ads9530 ↗Kurz gefasst
Fachartikel in der Zeitschrift Science, erschienen am 08.01.2026. Verfasst von Yinjun Jia und 22 weiteren Autor:innen. Beteiligte Einrichtung: Institute for AI Industry Research (AIR), Tsinghua University, Beijing, China. Hauptthemen laut MEDLINE: Deep Learning, Drug Discovery, Small Molecule Libraries. Die Arbeit wurde bisher 78-mal zitiert (Crossref) und nennt 117 Literaturangaben. Gefördert von: National Natural Science Foundation of China, National key R&D program of China.
Abstract (Originalsprache)
Recent breakthroughs in protein structure prediction have opened new avenues for genome-wide drug discovery, yet existing virtual screening methods remain computationally prohibitive. We present DrugCLIP, a contrastive learning framework that achieves ultrafast and accurate virtual screening, up to 10 million times faster than docking, while consistently outperforming various baselines on in silico benchmarks. In wet-lab validations, DrugCLIP achieved a 15% hit rate for norepinephrine transporter, and structures of two identified inhibitors were determined in complex with the target protein. For thyroid hormone receptor interactor 12, a target that lacks holo structures and small-molecule binders, DrugCLIP achieved a 17.5% hit rate using only AlphaFold2-predicted structures. Finally, we released GenomeScreenDB, an open-access database providing precomputed results for ~10,000 human proteins screened against 500 million compounds, pioneering a drug discovery paradigm in the post-AlphaFold era.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Substanzen
Autor:innen und Institutionen
- Institute for AI Industry Research (AIR), Tsinghua University, Beijing, China.
- School of Life Sciences, Tsinghua University, Beijing, China.
- IDG/McGovern Institute for Brain Research, Tsinghua University, Beijing, China.
- Tsinghua-Peking Center for Life Sciences, Beijing, China.
- State Key Laboratory of Membrane Biology, Tsinghua University, Beijing, China.
- Department of Computer Science and Technology, Tsinghua University, Beijing, China.
ORCID-Profile: Yinjun Jia, Bowen Gao, Jiaxin Tan, Jiqing Zheng, Xin Hong, Wenyu Zhu, Haichuan Tan, Yuan Xiao, Yanwen Huang, Zhiheng Deng, Xiangwei Wu, Yafei Yuan, Weiying Ma, Yaqin Zhang, Lei Liu, Chuangye Yan, Wei Zhang, Yanyan Lan
Förderung
- National Natural Science Foundation of China
- National key R&D program of China
Zitieren
APA
Jia, Y., Gao, B., Tan, J., Zheng, J., Hong, X., Zhu, W., Tan, H., Xiao, Y., Tan, L., Cai, H., Huang, Y., Deng, Z., Wu, X., Jin, Y., Yuan, Y., Tian, J., He, W., Ma, W., Zhang, Y., …, & Lan, Y. (2026). Deep contrastive learning enables genome-wide virtual screening. Science, 391(6781). https://doi.org/10.1126/science.ads9530
Vancouver
Jia Y, Gao B, Tan J, Zheng J, Hong X, Zhu W, et al. Deep contrastive learning enables genome-wide virtual screening. Science. 2026;391(6781). doi:10.1126/science.ads9530
BibTeX
@article{Jia2026,
author = {Jia, Yinjun and Gao, Bowen and Tan, Jiaxin and Zheng, Jiqing and Hong, Xin and Zhu, Wenyu and Tan, Haichuan and Xiao, Yuan and Tan, Liping and Cai, Hongyi and Huang, Yanwen and Deng, Zhiheng and Wu, Xiangwei and Jin, Yue and Yuan, Yafei and Tian, Jiekang and He, Wei and Ma, Weiying and Zhang, Yaqin and Liu, Lei and Yan, Chuangye and Zhang, Wei and Lan, Yanyan},
title = {Deep contrastive learning enables genome-wide virtual screening},
journal = {Science},
year = {2026},
volume = {391},
number = {6781},
doi = {10.1126/science.ads9530}
}
EUVIMED