F Foldworks
Foldworks · v0.1 preview

Design protein binders against any target
In an afternoon.

Pick a hotspot on your target. Foldworks diffuses de novo backbones with RFdiffusion, designs sequences with ProteinMPNN, and validates the complex with ESMFold — then ranks the candidates. Antibodies, small molecules, and fluorescent probes run against the same hotspot.

See what it does
~15 min median binder batch·5 site predictors·PDB / SDF / CSV output
Demo demo / demo · 5 design jobs/day · no signup
Cartoon rendering of SARS-CoV-2 main protease (PDB 6LU7) in complex with an inhibitor
PDB · 6LU7 · MOL*
6LU7· SARS-CoV-2 Mpro· 2.16 Å X-rayRCSB ↗
Built on
AlphaFoldRFdiffusionRFantibodyProteinMPNNAbMPNNBindCraftmberESMFoldBoltz-2Chai-1AutoDock VinaGNINADiffDock-LAiZynthFinderREINVENT 4ADMET-AIPoseBustersThermoMPNNPROSSDeepFRIPDBFixerOpenMMPyRosettaMeekoRDKitMol*UniProtRCSB PDBGTExOpen TargetsHuman Protein AtlasAlphaFoldRFdiffusionRFantibodyProteinMPNNAbMPNNBindCraftmberESMFoldBoltz-2Chai-1AutoDock VinaGNINADiffDock-LAiZynthFinderREINVENT 4ADMET-AIPoseBustersThermoMPNNPROSSDeepFRIPDBFixerOpenMMPyRosettaMeekoRDKitMol*UniProtRCSB PDBGTExOpen TargetsHuman Protein Atlas

Pipelines

Built around
binder design.

RFdiffusion-based protein binder design is the headline pipeline. Antibody CDRs, small-molecule docking, and fluorescent-probe screening all run against the same structure and the same hotspot — switch modes without re-uploading or re-defining sites. Jobs queue asynchronously with live progress.

★ Headline pipeline

Protein binder design

RFdiffusion or BindCraft → ProteinMPNN → ESMFold.

Pick a hotspot, or let the algorithm auto-detect it. Diffuse de novo miniprotein backbones with RFdiffusion — ~16% of ordered designs bind experimentally (Watson 2023). Or use BindCraft for AF2-hallucination design — Pacesa 2024 report ~10× higher wet-lab hit rate on short helical binders. ProteinMPNN designs sequences with ~52% native E. coli expression (vs ~9% for Rosetta). ESMFold validates the complex.

STEP 01
RFdiffusion
De novo backbone diffusion
~16% bind (Watson 2023)
STEP 02
ProteinMPNN
Sequence design
~52% E. coli expression
STEP 03
ESMFold
Complex validation
interface pLDDT · ipTM
STEP 04
Rank
8-design batch
~15 min on a 4090
RFdiffusion · BindCraft · ProteinMPNN · ESMFoldWatson 2023 ↗Pacesa 2024 ↗Dauparas 2022 ↗
Also supports

Antibody + nanobody design

Paired Fab CDRs (RFantibody) and de novo VHH binders (mber).

RFantibody + AbMPNN design CDR loops on an existing heavy+light framework — Bennett 2024 report ~17% wet-lab binding. mber generates de novo single-domain VHH / nanobody scaffolds with NanoBodyBuilder2 validation. Experimental — no developability scoring yet.

RFantibodyAbMPNNmber
Bennett 2024 ↗Manifold Bio 2025 ↗
Also supports

Small-molecule screening

Vina + GNINA + Boltz-2, with ML ADMET on every hit.

AutoDock Vina docks a curated drug-like library; GNINA reranks poses with a CNN trained on PDBbind. Boltz-2 affinity-rescores the top-5 hits. Every hit carries ADMET-AI predictions on 10 endpoints, plus RDKit Ro5 / Veber / PAINS triage.

AutoDock VinaGNINABoltz-2ADMET-AI
Trott 2010 ↗McNutt 2021 ↗Swanson 2024 ↗
Also supports

Sequence redesign + probes

ProteinMPNN sequence diversity + fluorophore screening.

ProteinMPNN redesigns sequences on an existing backbone — ~52% recovery, 6× the E. coli expression of Rosetta designs. The soluble-trained variant rescues designs that don't fold or ship. The same Vina engine screens a curated fluorescent-probe library and outputs Ex/Em wavelengths.

ProteinMPNNsoluble-MPNNFluorophore lib
Dauparas 2022 ↗
Platform fundamental

Binding-site prediction

Druggable pockets, PPI hotspots, antibody epitopes — mode-aware.

fpocket detects ~80% of known small-molecule sites; P2Rank's top-3 ranking covers ~90% of held-out pockets (COACH420). DoGSite3 adds druggability scoring, FreeSASA finds PPI hotspots, and DiscoTope-3 predicts antibody epitopes. For pipelines that support auto-detect, the algorithm picks its own site if you don't supply one.

fpocketP2RankDoGSite3DiscoTope-3

Built for chemists

Make every chemist a
computational chemist.

Foldworks runs the RFdiffusion, ProteinMPNN, Vina and AbMPNN pipelines that comp-chem groups normally assemble by hand — as one-click workflows with sensible defaults, citable methods, and plain-file output. If you can read a binding curve, you can run a binder design.

01

One-click pipelines.

Pick binder design, antibody CDR, small-molecule docking, or probe screening. Foldworks handles structure prep, parameter selection, ligand prep, and GPU queueing. No XML config files, no command-line flags, no licence dongles.

02

Defaults you can trust.

Every pipeline uses parameters drawn from the published methods — Watson 2023 for RFdiffusion, Dauparas 2022 for ProteinMPNN, Trott 2010 for Vina. The defaults work for the common cases; the methods are documented and citable.

03

Plain files, no lock-in.

Output is PDB / SDF / CSV with full provenance. Drop it into ChemDraw, Excel, your wet-lab tracking spreadsheet, or your existing pipeline. Nothing is hidden behind a proprietary format — and you can re-run anything later.

Data layer

Every target carries
its own context.

Before you design against a target, Foldworks fetches what's known about it — sequence, isoforms, subcellular location, tissue expression, tractability, predicted and experimental structures. From primary sources, with citations to the original entry, never through a black-box aggregator.

Access

Start free.
Scale on request.

The Researcher demo is permanently free for everyone. For higher quotas, shared workspaces, or self-hosted deployment behind your own VPC, write to info@foldworks.bio — pricing depends on volume and integration scope.

ResearcherFreeNo signup · demo / demo
  • 5 design jobs / day per IP
  • Unlimited structure viewer
  • All pipelines + ADMET triage
  • Plain PDB / SDF / CSV export
TeamContact usHigher quotas, shared workspaces
  • Per-seat or per-job billing
  • Projects + shared structures
  • Priority GPU queue
  • Email + Slack support
Email info@foldworks.bio →
Enterprise · self-hostedContact usIn your VPC, no data egress
  • Docker / Helm deployment
  • Custom libraries + private targets
  • SSO, audit logs, role-based access
  • Direct researcher support
Email info@foldworks.bio →

Methods

Every Foldworks pipeline is a thin orchestration layer over a published structural-biology method. Models, weights, and source code are documented and citable. Output is plain PDB / SDF / CSV — bring your own downstream toolchain.

Backbone diffusionRFdiffusion · RFantibody
VHH / nanobody designmber (Manifold Bio)
Structure predictionESMFold · AlphaFold DB · Boltz-2 · Chai-1
Small-mol dockingAutoDock Vina · GNINA · DiffDock-L
Pocket detectionfpocket · P2Rank · DoGSite3
ADMET / PK-PDADMET-AI (Chemprop) · RDKit triage
RetrosynthesisAiZynthFinder · rule-based templates
Function annotationDeepFRI (GO terms, EC numbers)
Structure prepPDBFixer · OpenMM
Hallucination designBindCraft (AF2 + ColabDesign)
Sequence designProteinMPNN · AbMPNN · soluble-MPNN
Affinity predictionBoltz-2 · Chai-1 (AF3-class)
De novo small-molREINVENT 4 (RL-based)
Epitope predictionFreeSASA · DiscoTope-3
Pose validationPoseBusters (~20 geometry checks)
Stability predictionThermoMPNN · PROSS
Post-design refinementPyRosetta FastRelax
VisualizationMol* (RCSB)