Novel Molecules Generation Using Graph Generative Adversarial Networks
Molecular Informatics · 2026-05 · 45 (5)
DOI: 10.1002/minf.70031 ↗Kurz gefasst
Fachartikel in der Zeitschrift Molecular Informatics, online erschienen am 10.05.2026. Verfasst von Amodini A P und 2 weiteren Autor:innen. Beteiligte Einrichtung: Computer Science and Engineering Group, Indian Institute of Information Technology, Sri City, India. Hauptthemen laut MEDLINE: Drug Discovery. Die Arbeit nennt 24 Literaturangaben. Gefördert von: Anusandhan National Research Foundation.
Abstract (Originalsprache)
De novo molecular generation remains a central challenge in drug discovery due to the vastness of chemical space and the difficulty of generating molecules that are simultaneously valid, diverse, and property‐aware. While graph‐based generative models have improved structural validity over sequence‐based approaches, many existing methods suffer from mode collapse, limited controllability, and unstable reinforcement learning dynamics. In this work, we propose NovMolG‐GAN , a graph‐based generative adversarial framework that integrates graph attention mechanisms, a learnable reward network, mode‐seeking regularization, and proximal policy optimization‐based reinforcement learning to address these limitations. The proposed model enables stable training while explicitly encouraging molecular validity, novelty, diversity, and drug‐likeness. Experiments on the ChEMBL‐35 dataset demonstrate that NovMolG‐GAN achieves high validity (99.6%), high novelty (99.4%) strong uniqueness (88.5%), and competitive quantitative drug‐likeness scores under unconstrained generation. Furthermore, a conditional generation ablation incorporating synthetic accessibility as an additional reward illustrates the controllability of the framework and its ability to steer generation toward synthesis‐feasible regions of chemical space. These results highlight the potential of NovMolG‐GAN as a flexible and extensible foundation for goal‐directed molecular design.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Computer Science and Engineering Group, Indian Institute of Information Technology, Sri City, India.
- Department of Computer Science and Engineering, Indian Institute of Information Technology Design and Manufacturing Kurnool, Kurnool, India.
ORCID-Profile: Chandra Mohan Dasari
Förderung
- Anusandhan National Research Foundation
Zitieren
APA
A P, A., Dasari, C. M., & Amilpur, S. (2026). Novel Molecules Generation Using Graph Generative Adversarial Networks. Molecular Informatics, 45(5). https://doi.org/10.1002/minf.70031
Vancouver
A P A, Dasari CM, Amilpur S. Novel Molecules Generation Using Graph Generative Adversarial Networks. Molecular Informatics. 2026;45(5). doi:10.1002/minf.70031
BibTeX
@article{AP2026,
author = {A P, Amodini and Dasari, Chandra Mohan and Amilpur, Santhosh},
title = {Novel Molecules Generation Using Graph Generative Adversarial Networks},
journal = {Molecular Informatics},
year = {2026},
volume = {45},
number = {5},
doi = {10.1002/minf.70031}
}
EUVIMED