bio.rodeo
ModelsOrganizationsProvidersLeaderboardAboutSign in
bio.rodeo

The authoritative source for evaluating biological foundation models. No hype, just honest analysis.

Categories
  • DNA & Gene
  • RNA
  • Protein
  • Small molecule
  • Single-cell
  • Spatial omics
  • Pathology
  • Imaging
  • Metabolomics
  • Biosignals
  • Language model
bio.rodeoModelsOrganizationsProvidersLeaderboardAboutFAQSubmit a modelContact
© 2026 Pulsatance. All rights reserved. ~
Built by Pulsatance
models / protein / dynamicflow
ProteinSmall molecule
Tsinghua UniversityUniversity of Chinese Academy of SciencesChinese Academy of SciencesPeking UniversityUniversity of Illinois Urbana-ChampaignGriffith UniversityStoneWiseReleased March 2025

DynamicFlow

Full-atom flow matching model that generates a ligand and the induced-fit holo pocket together, starting from an apo binding site.

The short version

  • —Designs a ligand for a target whose only available structure is the unbound apo form
  • —Emits the induced-fit pocket conformation alongside the molecule in a single pass
  • —Refined pockets drop into rigid generators, lifting IPDiff Vina from -7.55 to -11.04
  • —One flow spans Euclidean positions, SO(3) frames, side-chain tori, and discrete bond types
9Openness3Citations

Where to run it

No providers recorded yet. Browse all providers

Generative structure-based drug design almost always assumes the pocket holds still. A model is handed a fixed binding site and asked to invent a molecule for it, which ignores induced fit: side chains and backbone rearrange as a ligand binds, and the holo conformation often differs enough from the apo one that molecules designed against the latter do not fit. Molecular dynamics can sample the missing conformations, but crossing the energy barriers between them is too expensive for a design loop.

DynamicFlow, accepted as a poster at ICLR 2025, learns the transition instead. It comes from a collaboration led by Tsinghua University with the Chinese Academy of Sciences, Peking University, and the drug discovery company StoneWise. The model is a full-atom flow taking an apo pocket plus noise-sampled ligand atoms at time zero and integrating them to a holo pocket plus a bound 3D molecule at time one, generating ligand chemistry, ligand geometry, and pocket motion in one pass rather than in stages.

Flow matching had already been applied to protein conformational change on its own — as in AlphaFlow, which samples ensembles for a protein in isolation. DynamicFlow's departure is coupling that motion to the molecule causing it, which yields a second usable product: the predicted pocket conformations can be handed to conventional rigid-pocket generators as better inputs than the apo structure.

#Key Features

  • Joint pocket and ligand generation: One generative process outputs residue frames, side-chain torsions, ligand atom positions, atom types, and bond types, so molecule and accommodating conformation are consistent by construction.
  • Multi-manifold flow: Continuous variables follow conditional flow matching on their natural manifolds — Euclidean space for atom positions, SO(3) geodesics for residue rotations, tori for torsion angles — while discrete atom and bond types follow continuous-time Markov chains.
  • Stochastic variant: DynamicFlow-SDE extends the ODE formulation by injecting Gaussian noise along the conditional probability path, trading geometric precision for broader coverage of holo states.
  • Pocket refinement for existing tools: The generated holo-like conformations substitute for apo inputs to rigid-pocket generators, improving their designs without changing those models.

#Technical Details

Training data derives from MISATO, which supplies 8 ns of recorded molecular dynamics trajectory for 16,972 protein-ligand complexes taken from PDBbind. Removing oligopeptide ligands leaves 12,695 complexes; pockets are residues within 7 Å of the ligand across 100 trajectory frames, clustered with GROMACS at a 1 Å heavy-atom RMSD threshold and cut to the top ten clusters per complex. Apo counterparts come from AlphaFold Protein Structure Database entries mapped through UniProt, yielding 7,528 aligned apo pockets. Filtering on average ligand RMSD below 3 Å gives the final set: 5,692 apo/holo pairs carrying 46,235 holo-ligand conformations. An atom-level SE(3)-equivariant graph network processes ligand and protein atoms together and pools its pocket features to residue level for an invariant point attention Transformer predicting frames and torsions; the model totals 15.9M parameters and trains on seven weighted losses including a protein-ligand interaction term.

Evaluation uses 50 held-out targets, generating 100 molecules each and scoring the best by AutoDock Vina in minimize mode, with every method given the apo pocket. DynamicFlow-SDE reports a Vina score of -7.65 with QED 0.53 and 52.5% of designs beating the reference ligand; the ODE variant, -7.28 and 51.0%. Reference ligands average -7.84. Baselines on the same apo inputs reach -5.50 for Pocket2Mol, -5.09 for TargetDiff, and -7.55 for IPDiff. Substituting DynamicFlow's refined pockets lifts TargetDiff to -9.00 and its high-affinity rate from 44.0% to 61.2%, and IPDiff to -11.04 and 51.0% to 83.7%. Generating ten molecules takes 35 seconds at 100 function evaluations, against 156 seconds for TargetDiff and 980 for Pocket2Mol.

#Applications

The model is built for targets where the informative structure is the one that does not exist yet: a protein solved only in its apo form, a cryptic pocket that opens on binding, or a site where docking against the unbound conformation has repeatedly disappointed. It offers two entry points — sampling ligands directly, or using it as a preprocessing step that turns an apo structure into a holo-like one before an established design or docking pipeline. The second mode is lower-friction, since it slots into workflows already in place.

#Impact

DynamicFlow reframes structure-based design as generating a complex rather than a ligand for a fixed receptor, and the pocket-refinement result is the strongest evidence the reframing pays: two unrelated generators improve substantially on inputs the model produces, a transferable gain rather than a self-contained benchmark win. The bounds are clear. Evaluation covers 50 targets scored entirely in silico, with Vina, QED, and synthetic accessibility standing in for properties only assays can settle; training conformations come from 8 ns simulations, short relative to large conformational transitions; and no code or trained weights have been released, so the results cannot be independently reproduced.

At a glance

Parameters
15.9 Million
Released
March 2025
Category
Protein
Organizations
Tsinghua University / University of Chinese Academy of Sciences / Chinese Academy of Sciences / Peking University / University of Illinois Urbana-Champaign / Griffith University / StoneWise

Related models

  • DrugFlow

    EPFL / VantAI / University of Oxford / Aithyra

  • PAFlow

    Shanghai Jiao Tong University / Guangdong Laboratory of Artificial Intelligence and Digital Economy (Shenzhen)

  • IDFlow

    KTH Royal Institute of Technology / Science for Life Laboratory / Stockholm University / The Alan Turing Institute / Max Planck Institute for Polymer Research

  • MolPIF

    Lingang Laboratory / Shanghai Institute of Materia Medica / ShanghaiTech University / Fudan University / Shanghai Jiao Tong University

  • FlexiFlow

    AstraZeneca / KTH Royal Institute of Technology / Chalmers University of Technology / Science for Life Laboratory / Stockholm University

  • FusedBFN

    Shanghai Jiao Tong University / Guangdong Laboratory of Artificial Intelligence and Digital Economy (Shenzhen)

Links

Research Paper

Tags

flow_matchinggenerativegraph_neural_networkmolecule_generation

Something wrong?

Much of this page is generated or calculated automatically. Flag anything that looks off and we will re-run it.