RetroGEF: Dynamic Graph Edit Flow for Single-Step Retrosynthesis

Abstract

Retrosynthesis enables the discovery of viable synthetic routes to target molecules. It plays a central role in modern drug discovery and materials design. Retrosynthesis involves molecular graph transformations that can change both connectivity and graph size. These transformations may introduce reactant components absent from the target while revising the product-derived structure. To model these transformations, we propose RetroGEF, a flow-based generative model for single-step retrosynthesis. Starting from the target molecule, it constructs possible reactants by adding atoms and changing bonds in the molecular graph. RetroGEF models molecular transformations and changes in graph size within the same generative process, rather than relying on a fixed-size graph canvas. It learns this process directly from product--reactant pairs without requiring a prescribed edit order. Experiments on representative retrosynthesis benchmarks demonstrate that RetroGEF achieves state-of-the-art performance.

Multimodal Brain Graph Analysis for Uncovering Path-Based Biomarkers in Schizophrenia: Integrating Static and Dynamic Approaches.
Haleh Falakshahi, Hooman Rokham, Daniel H Mathalon +7
Human brain mapping1 Oct 2026Peer-reviewedCC BYFull textNeuroscienceEurope PMC
Legal judgment prediction via dynamic task graph and keyword-aware contrastive learning
Jiawei Li, Shengjia Zhang
PLOS One10 Sep 2026Peer-reviewedCC BY 4.0Full textNeurosciencePLOS
Dynamic hypernode injection and graph attention fusion for multimodal sentiment analysis of behavioral signals
Wenfei Cao, Jian Xu, Yinghao Li +3
PLOS One28 Sep 2026Peer-reviewedCC BY 4.0Full textBioinformatics & computational biology NeurosciencePLOS
Modular constraint-based joint dynamic functional brain graph learning model for identifying brain disorders
Peiming Xu, Yaru Li, Wei Si +2
PeerJ30 Sep 2026Peer-reviewedCC BY 4.0Full textNeuroscienceCrossref
Dynamic regulation of single-cell contractile heterogeneity in the human trabecular meshwork
L. SANCHEZ, A. C. Jubintoro, I. Pushkarsky +3
bioRxiv6 Nov 2026PreprintAll rights reservedFull textCell biologybioRxiv

Cite

BibTeX RIS

BibTeX

@misc{song2026retrogef,
  title = {{RetroGEF: Dynamic Graph Edit Flow for Single-Step Retrosynthesis}},
  author = {Xiaozhuang Song and Xuemin Chen and Xinjian Zhao and Yaoyao Xu and Tianshu Yu},
  howpublished = {arXiv},
  year = {2026},
  month = sep,
  doi = {10.48550/arxiv.2609.38484},
  eprint = {2609.38484},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2609.38484},
}

RIS

TY  - UNPB
TI  - RetroGEF: Dynamic Graph Edit Flow for Single-Step Retrosynthesis
AU  - Xiaozhuang Song
AU  - Xuemin Chen
AU  - Xinjian Zhao
AU  - Yaoyao Xu
AU  - Tianshu Yu
PY  - 2026
DA  - 2026/09/29
PB  - arXiv
DO  - 10.48550/arxiv.2609.38484
UR  - https://arxiv.org/abs/2609.38484
AB  - Retrosynthesis enables the discovery of viable synthetic routes to target molecules. It plays a central role in modern drug discovery and materials design. Retrosynthesis involves molecular graph transformations that can change both connectivity and graph size. These transformations may introduce reactant components absent from the target while revising the product-derived structure. To model these transformations, we propose RetroGEF, a flow-based generative model for single-step retrosynthesis. Starting from the target molecule, it constructs possible reactants by adding atoms and changing bonds in the molecular graph. RetroGEF models molecular transformations and changes in graph size within the same generative process, rather than relying on a fixed-size graph canvas. It learns this process directly from product--reactant pairs without requiring a prescribed edit order. Experiments on representative retrosynthesis benchmarks demonstrate that RetroGEF achieves state-of-the-art performance.
ER  -

Abstract

Related

Cite