Skip to content
DraconiticResearch
Search Sources API
203,203 studies31 sourcesUpdated 41 min ago
Peer-reviewedCC BYNature Computational Science

FLOWR: flow matching for structure-aware de novo, interaction- and fragment-based ligand generation

  • Julian Cremer1
  • Ross Irwin2,3
  • Alessandro Tibo4
  • Jon Paul Janet4
  • Simon Olsson5
  • Djork-Arné Clevert6
6 affiliations
  1. 1Machine Learning and Computational Sciences, Pfizer Worldwide R&D, Berlin, Germany. [email protected]
  2. 2Department of Computer Science and Engineering, Chalmers University of Technology and University of Gothenburg, Gothenburg, Sweden. [email protected]
  3. 3Molecular AI, Discovery Sciences, R&D, AstraZeneca, Gothenburg, Sweden. [email protected]
  4. 4Molecular AI, Discovery Sciences, R&D, AstraZeneca, Gothenburg, Sweden
  5. 5Department of Computer Science and Engineering, Chalmers University of Technology and University of Gothenburg, Gothenburg, Sweden
  6. 6Machine Learning and Computational Sciences, Pfizer Worldwide R&D, Berlin, Germany
Open at sourcePDFDOICite

Abstract

Here we introduce FLOWR, a structure-based framework for the generation and optimization of three-dimensional ligands. FLOWR integrates continuous and categorical flow matching with equivariant optimal transport, enhanced by an efficient protein pocket conditioning. Alongside FLOWR, we present SPINDR, a curated dataset comprising ligand-pocket cocrystal complexes specifically designed to address existing data quality issues. Empirical evaluations demonstrate that FLOWR surpasses current state-of-the-art diffusion- and flow-based methods in terms of PoseBusters-validity, pose accuracy and interaction recovery, while offering an inference speed-up, achieving up to 70-fold faster performance. In addition, we introduce FLOWR.MULTI, a highly accurate multi-purpose model allowing for the targeted sampling of ligands that adhere to predefined interaction profiles and chemical substructures for fragment-based design without the need of retraining or any resampling strategies. Collectively, our results indicate that FLOWR and FLOWR.MULTI represent an advancement in artificial intelligence-driven structure-based drug design, substantially enhancing the reliability and applicability of de novo, interaction- and fragment-based ligand generation in real-world drug discovery settings.

Related

  1. FLOWR.ROOT – A flow matching-based foundation model for joint multi-purpose structure-aware 3D ligand generation and affinity prediction
    Julian Cremer, Tuan Le, Mohammad Mehdi Ghahremanpour +3
    Nature CommunicationsPeer-reviewedCC BYFull text9 citedMedicinal chemistry & drug discoveryOpenAlex
  2. Comprehensive Quantum Chemical Interaction Analysis of Whole Protein–Ligand Structures in BindingDB Using the Fragment Molecular Orbital Method
    Shuhei Miyakawa, Naoko Konami, Ruri Mihata +10
    ChemRxivPreprintCC BYFull textGeneral chemistryChemRxiv
  3. Curvature-Aware Flow Matching for Molecular Structure Generation
    Samir Darouich, Juliane Geisler, Jacob W. Toney +4
    arXivPreprintCC BY 4.0Full textPhysical chemistryarXiv
  4. Discovery of MYC/NSD3S Protein–Protein Interaction Negative Modulators with a Live Cell Nanoluciferase Protein-Fragment Complementation uHTS Assay
    Sebastián Vera-Molina, Diego Acosta-Yanez, Ebonny Olivera +6
    ACS OmegaPeer-reviewedCC BY-NC-ND 4.0Full textGeneral chemistryCrossrefEurope PMC
  5. Quantum Graph-Theoretic Docking of Multiple Ligands Targeting CMG Helicase: Analysis of Interaction Patterns and Ideal Ligand Composition
    Marina Gavrilovskaia, Katarzyna Mizgalska, Timothy H. Tran +6
    ChemRxivPreprintCC BYFull textGeneral chemistryChemRxiv

Cite

BibTeXRIS
BibTeX
@article{cremer2026flowr,
  title = {{FLOWR: flow matching for structure-aware de novo, interaction- and fragment-based ligand generation}},
  author = {Julian Cremer and Ross Irwin and Alessandro Tibo and Jon Paul Janet and Simon Olsson and Djork-Arné Clevert},
  journal = {Nature Computational Science},
  year = {2026},
  month = may,
  doi = {10.1038/s43588-026-00998-8},
  url = {https://doi.org/10.1038/s43588-026-00998-8},
  note = {CC BY},
}
RIS
TY  - JOUR
TI  - FLOWR: flow matching for structure-aware de novo, interaction- and fragment-based ligand generation
AU  - Julian Cremer
AU  - Ross Irwin
AU  - Alessandro Tibo
AU  - Jon Paul Janet
AU  - Simon Olsson
AU  - Djork-Arné Clevert
PY  - 2026
DA  - 2026/05/28
JO  - Nature Computational Science
DO  - 10.1038/s43588-026-00998-8
UR  - https://doi.org/10.1038/s43588-026-00998-8
AB  - Here we introduce FLOWR, a structure-based framework for the generation and optimization of three-dimensional ligands. FLOWR integrates continuous and categorical flow matching with equivariant optimal transport, enhanced by an efficient protein pocket conditioning. Alongside FLOWR, we present SPINDR, a curated dataset comprising ligand-pocket cocrystal complexes specifically designed to address existing data quality issues. Empirical evaluations demonstrate that FLOWR surpasses current state-of-the-art diffusion- and flow-based methods in terms of PoseBusters-validity, pose accuracy and interaction recovery, while offering an inference speed-up, achieving up to 70-fold faster performance. In addition, we introduce FLOWR.MULTI, a highly accurate multi-purpose model allowing for the targeted sampling of ligands that adhere to predefined interaction profiles and chemical substructures for fragment-based design without the need of retraining or any resampling strategies. Collectively, our results indicate that FLOWR and FLOWR.MULTI represent an advancement in artificial intelligence-driven structure-based drug design, substantially enhancing the reliability and applicability of de novo, interaction- and fragment-based ligand generation in real-world drug discovery settings.
SN  - 2662-8457
ER  -