De Novo Molecular Design via Shape-Constrained Diffusion Models
Journal of Chemical Information and Modeling · 27.04.2026 · 66 (8) · 4592-4606
DOI: 10.1021/acs.jcim.6c00044 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Chemical Information and Modeling, online erschienen am 13.04.2026. Verfasst von Bohao Li und 13 weiteren Autor:innen. Beteiligte Einrichtung: School of Computer Science and Engineering, Sun Yat-Sen University, Guangzhou 510006, China. Hauptthemen laut MEDLINE: Drug Design, Models, Molecular. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 39 Literaturangaben. Gefördert von: Pearl River Recruitment Program of Talents, Guangzhou National Laboratory.
Abstract (Originalsprache)
Generative models that preserve 3D shape similarity while enabling structural novelty remain a central challenge in de novo molecular design. We present Diff-Shape, a diffusion-based framework that couples a pretrained unconditional 3D generator with a Graph ControlNet constraint that ingests a 3D reference shape. By integrating substructure inpainting, Diff-Shape enables controllable, shape-guided design across tasks such as scaffold hopping, scaffold decoration, and linker generation. On comprehensive benchmarks, Diff-Shape achieves substantially higher rates of 3D shape fidelity than state-of-the-art methods, while maintaining low 2D graph similarity. These improvements are robust across novelty thresholds, noise levels, and diverse references. Finally, case studies on Kirsten rat sarcoma virus (KRAS) G12D and epidermal growth factor receptor L858R/T790M/C797S demonstrate that several designed candidates were synthesized and exhibit nanomolar biochemical potency, underscoring the method's translational relevance. Together, these results position Diff-Shape as a powerful and generalizable approach for shape-constrained molecular generation.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Substanzen
Autor:innen und Institutionen
- School of Computer Science and Engineering, Sun Yat-Sen University, Guangzhou 510006, China.
- Guangzhou National Laboratory, Guangzhou 510005, China.
- School of Pharmaceutical Science, Sun Yat-sen University, Guangzhou 510006, China.
- Guangzhou Medical University, Guangzhou 511495, China.
ORCID-Profile: Bohao Li, Jinping Lei, Yuedong Yang, Jinsai Shang, Hongming Chen
Förderung
- Pearl River Recruitment Program of Talents
- Guangzhou National Laboratory
Zitieren
APA
Li, B., Wu, X., Cao, Y., Lin, J., Zhong, J., Chen, H., Lu, Y., Tang, M., Lei, J., Ran, T., Yang, Y., Xu, M., Shang, J., & Chen, H. (2026). De Novo Molecular Design via Shape-Constrained Diffusion Models. Journal of Chemical Information and Modeling, 66(8), 4592-4606. https://doi.org/10.1021/acs.jcim.6c00044
Vancouver
Li B, Wu X, Cao Y, Lin J, Zhong J, Chen H, et al. De Novo Molecular Design via Shape-Constrained Diffusion Models. Journal of Chemical Information and Modeling. 2026;66(8):4592-4606. doi:10.1021/acs.jcim.6c00044
BibTeX
@article{Li2026,
author = {Li, Bohao and Wu, Xinyu and Cao, Yu and Lin, Jie and Zhong, Jingpeng and Chen, Hua and Lu, Yongzhi and Tang, Miru and Lei, Jinping and Ran, Ting and Yang, Yuedong and Xu, Mingyuan and Shang, Jinsai and Chen, Hongming},
title = {De Novo Molecular Design via Shape-Constrained Diffusion Models},
journal = {Journal of Chemical Information and Modeling},
year = {2026},
volume = {66},
number = {8},
pages = {4592--4606},
doi = {10.1021/acs.jcim.6c00044}
}
EUVIMED