FreeBindCraft

License: FreeBindCraft uses MIT for code and CC-BY-4.0 for model weights and may require explicit attribution when utilized. Please refer to the code license and model weights license for full terms.

This toolkit is open source. Any third-party models, product names, or trademarks referenced are the property of their respective owners, and Proto is not affiliated with them.

GitHub GitHub Publication Publication Cite Cite Tool Source Tool Source Open as Notebook Open as Notebook Open on Proto Open on Proto

cytokineking/FreeBindCraft

View repo

One-shot design of functional protein binders with BindCraft

Martin Pacesa, Lennart Nickel, … Bruno E. Correia

Nature (2025)

Read paper

@article{pacesa2025bindcraft,
  title={One-shot design of functional protein binders with BindCraft},
  author={Pacesa, Martin and Nickel, Lennart and Schellhaas, Christian and Schmidt, Joseph and Pyatova, Ekaterina and Kissling, Lucas and Barendse, Patrick and Choudhury, Jagrity and Kapoor, Srajan and Alcaraz-Serna, Ana and Cho, Yehlin and Ghamary, Kourosh H. and Vinu{\'e}, Laura and Yachnin, Brahm J. and Wollacott, Andrew M. and Buckley, Stephen and Westphal, Adrie H. and Lindhoud, Simon and Georgeon, Sandrine and Goverde, Casper A. and Hatzopoulos, Georgios N. and G{\"o}nczy, Pierre and Muller, Yannick D. and Schwank, Gerald and Swarts, Daan C. and Vecchio, Alex J. and Schneider, Bernard L. and Ovchinnikov, Sergey and Correia, Bruno E.},
  journal={Nature},
  volume={646},
  number={8084},
  pages={483--492},
  year={2025},
  month={Aug},
  publisher={Springer Science and Business Media LLC},
  doi={10.1038/s41586-025-09429-6},
  issn={1476-4687}
}

@misc{freebindcraft2025,
  title={FreeBindCraft: BindCraft with optional, PyRosetta-free scoring and relaxation},
  author={{cytokineking} and {FreeBindCraft contributors}},
  year={2025},
  howpublished={\url{https://github.com/cytokineking/FreeBindCraft}},
  note={MIT-licensed fork of BindCraft replacing PyRosetta with OpenMM, FreeSASA, and sc-rs}
}

Copy citation

evo-design/proto-tools/proto_tools/tools/binder_design/freebindcraft

View source

Open Notebook

Open notebook

FreeBindCraft Binder Design

Function	Description
`run_freebindcraft_design()`	PyRosetta-free de novo binder design: AlphaFold2 hallucination + ProteinMPNN refinement + AlphaFo… (GPU)	Docs Source

Background

BindCraft (Pacesa et al., 2025) addresses the problem of generating protein binders against a target without high-throughput experimental screening or curated structural templates, reporting experimental success rates of 10 to 100 percent across diverse targets. FreeBindCraft preserves the identical AlphaFold2-hallucination and ProteinMPNN design logic — the hallucination losses, MPNN sampling, and AF2 validation are unchanged — while swapping out the proprietary scoring backend. The pipeline chains four stages per design trajectory. First, an AlphaFold2 hallucination step initialises a binder of randomly sampled length adjacent to the frozen target and optimises the binder logits by gradient descent against a weighted sum of structural losses (per-residue pLDDT, intra-binder and inter-chain PAE, intra-binder and interface contact counts, interface pTM, a helicity bias, and a radius-of-gyration term). Second, the hallucinated backbone is handed to ProteinMPNN (Dauparas et al., 2022), which samples foldable sequences while optionally holding interface residues fixed. Third, each refined complex is re-predicted from scratch with AlphaFold2 multimer (Jumper et al., 2021) as an independent validation. Fourth, the validated complex is relaxed and scored — where BindCraft uses PyRosetta, FreeBindCraft uses an OpenMM-based relaxation protocol (PDBFixer cleanup, FASPR side-chain packing, implicit solvation) and computes interface metrics with FreeSASA, Biopython, and the sc-rs shape-complementarity binary.

PyRosetta-free metrics

FreeBindCraft computes the AlphaFold2 confidence metrics (pLDDT, pTM, interface pTM, ipSAE, PAE) and the geometry-based interface metrics — shape complementarity (sc-rs), buried surface area and SASA fractions (FreeSASA), interface-residue counts, secondary-structure composition, hotspot/target/binder RMSDs, and pre- and post-relaxation clash counts (geometric, Biopython) — for real. Metrics that depend on Rosetta’s energy function and lack an open-source equivalent (interface binding energy dG, dG/dSASA, Binder_Energy_Score, PackStat, and hydrogen-bond counts) are emitted upstream as placeholders only to satisfy default filters; this toolkit does not surface them, so every metric returned by freebindcraft-design is a real measurement.

Learning Resources

cytokineking/FreeBindCraft. The PyRosetta-free fork, its --no-pyrosetta install/runtime flag, and the technical overview of the open-source scoring replacements.
martinpacesa/BindCraft (Correia Lab, EPFL). The upstream BindCraft repository, command-line interface, and reference filter configurations.

Tools

FreeBindCraft Binder Design (`freebindcraft-design`)

Designs one or more de novo protein binders against a user-supplied target. The tool takes a target structure together with the target chain identifiers, an optional hotspot residue list, and a binder length range, and runs the FreeBindCraft pipeline until either the requested number of accepted designs has been produced or the configured trajectory limit has been reached. The output carries each accepted binder as an amino-acid sequence, an OpenMM-relaxed target-binder complex Structure with per-residue pLDDT in the B-factor column, and the per-design PyRosetta-free metrics used by the filter check.

API Reference

Source

Input: FreeBindCraftInput

target_pdb

Structure

required

Target structure. Accepts a file path, raw PDB/CIF content string, Structure object, or a dict in the shape produced by Structure.model_dump(mode='json').

Show Structure

structure

string

required

Raw structure content in PDB or CIF format.

structure_format

string

Format of the content string (auto-detected if omitted).

b_factor_type

BFactorType

default:"unspecified"

What the B-factor column represents.

source

string

Optional source identifier (filepath or tool name).

metrics

Metrics

Associated metrics (e.g., pLDDT, pTM scores, per-chain lists, pairwise matrices). None values are stripped at construction.

target_chain

string

default:"A"

Chain ID(s) of the frozen target (comma-separated for multi-chain). Maps to BindCraft’s chains.

target_hotspot_residues

string

Comma-separated 1-indexed residue positions on the target that the binder must contact. Supports ranges (e.g. "1-10,56,78"). None or empty = unrestricted.

binder_lengths

array

default:"[65, 150]"

(min, max) binder length range. Maps to BindCraft’s lengths.

binder_name

string

default:"binder"

Project identifier — used as a prefix in output filenames.

number_of_final_designs

integer

default:"100"

Target accepted-design count. The pipeline stops after reaching this count or after max_trajectories attempts (whichever comes first).

Source

Config: FreeBindCraftConfig

design_algorithm

enum

default:"4stage"

Hallucination algorithm. Drives which iteration-count fields below are actually consumed (see each field’s description). UpstreamAvailable options: 2stage, 3stage, 4stage, greedy, mcmc

use_multimer_design

boolean

default:"True"

Use AF2 multimer parameters during hallucination. Every upstream preset uses multimer.

omit_AAs

string

default:"C"

Amino acids to ban during design (no separator). Upstream default: "C".

force_reject_AA

boolean

default:"False"

Reject any design containing omit_AAs.

soft_iterations

integer

default:"75"

Soft-stage iterations. Used by 2stage/3stage/4stage.

temporary_iterations

integer

default:"45"

Temporary-stage iterations. Used by 3stage/4stage.

hard_iterations

integer

default:"5"

Hard-stage iterations. Used by 3stage/4stage.

greedy_iterations

integer

default:"15"

Greedy/MCMC iterations. Used by 2stage/4stage/greedy/mcmc.

greedy_percentage

number

default:"1.0"

Greedy/MCMC mutation rate as % of binder length.

weights_plddt

number

default:"0.1"

pLDDT loss weight.

weights_pae_intra

number

default:"0.4"

Intra-chain PAE loss weight.

weights_pae_inter

number

default:"0.1"

Inter-chain (interface) PAE loss weight.

weights_con_intra

number

default:"1.0"

Intra-chain contact loss weight.

weights_con_inter

number

default:"1.0"

Inter-chain (interface) contact loss weight.

weights_helicity

number

default:"-0.3"

Helicity bias weight (negative discourages helices).

weights_iptm

number

default:"0.05"

Interface pTM loss weight (only used when use_i_ptm_loss=True).

weights_rg

number

default:"0.3"

Radius-of-gyration loss weight (only used when use_rg_loss=True).

weights_termini_loss

number

default:"0.1"

N-/C-termini distance loss weight (only used when use_termini_distance_loss=True).

random_helicity

boolean

default:"False"

Randomize the sign of weights_helicity per trajectory.

use_i_ptm_loss

boolean

default:"True"

Enable interface pTM loss.

use_rg_loss

boolean

default:"True"

Enable radius-of-gyration loss.

use_termini_distance_loss

boolean

default:"False"

Enable termini-distance loss.

rm_template_seq_design

boolean

default:"False"

Mask target template sequence during hallucination.

rm_template_seq_predict

boolean

default:"False"

Mask target template sequence during validation.

predict_initial_guess

boolean

default:"False"

Use the trajectory structure as AF2’s initial guess.

predict_bigbang

boolean

default:"False"

Use AF2’s “Big Bang” recycle initialisation.

enable_mpnn

boolean

default:"True"

Run ProteinMPNN sequence refinement after each accepted trajectory. When False, num_seqs / max_mpnn_sequences are inert.

mpnn_fix_interface

boolean

default:"True"

Fix interface residues during MPNN redesign.

num_seqs

integer

default:"20"

Number of MPNN sequences to sample per trajectory.

max_mpnn_sequences

integer

default:"2"

Max MPNN sequences to validate per trajectory.

optimise_beta

boolean

default:"True"

4stage-only — increase recycles + iterations mid-trajectory when the soft-stage output is beta-heavy.

max_trajectories

integer | boolean

default:"False"

Max hallucination trajectories before stopping. False (upstream default) = unlimited; positive int = cap.

enable_rejection_check

boolean

default:"True"

Enable rolling acceptance-rate monitoring (stops the run if it stalls).

acceptance_rate

number

default:"0.01"

Minimum design acceptance rate to keep running.

start_monitoring

integer

default:"600"

Trajectory count before acceptance-rate monitoring starts.

filter_overrides

Dict[string, any]

Per-metric threshold overrides merged on top of the upstream default filters at dispatch time. Keys are upstream metric names (e.g. "Average_pLDDT"); values are upstream filter dicts (e.g. {"threshold": 0.85, "higher": True}).

verbose

integer

default:"0"

Verbosity level (0=quiet, 1=info, 2=debug, 3=raw subprocess stderr). True is coerced to 1 and False to 0.

device

string

default:"cuda"

Device to run the tool on.

timeout

integer

Maximum execution time in seconds. None (default) waits indefinitely.

seed

integer

Random seed. When set, tools run reproducibly up to small GPU float noise (see BaseToolOutput.approx_equal), and the seed participates in cache keys. When None, cacheable seed-sensitive tools skip cache until seeded.

Source

Output: FreeBindCraftOutput

designs

List[FreeBindCraftDesign]

Accepted binder designs (length is at most FreeBindCraftInput.number_of_final_designs).

Show FreeBindCraftDesign

design_name

string

required

Unique design identifier emitted by upstream (e.g. "binder_l60_s12345_mpnn3").

binder_sequence

string

required

Designed binder amino-acid sequence (1-letter codes).

structure

Structure

required

OpenMM-relaxed target+binder complex; B-factors are pLDDT on the 0-100 PDB scale (b_factor_type=PLDDT).

metrics

FreeBindCraftMetrics

required

Per-design averaged metrics that the filter check evaluates against.

seed

integer

required

Random seed of the trajectory that produced this design.

interface_aas

Dict[string, integer]

Amino-acid composition at the binder-target interface.

interface_residues

List[integer]

1-indexed binder residue positions at the interface.

n_trajectories_run

integer

Total trajectories attempted before stopping (success or hitting max_trajectories).

n_designs_accepted

integer

Designs that passed all filters (equals len(designs)).

Metrics

Metric	Type	Range
`avg_plddt`	float	0.0 to 1.0
`avg_ptm`	float	0.0 to 1.0
`avg_iptm`	float	0.0 to 1.0
`avg_pae`	float	≥ 0.0
`avg_ipae`	float	≥ 0.0
`avg_ipsae`	float	0.0 to 1.0
`avg_iplddt`	float	0.0 to 1.0
`avg_ss_plddt`	float	0.0 to 1.0
`avg_binder_plddt`	float	0.0 to 1.0
`avg_binder_ptm`	float	0.0 to 1.0
`avg_binder_pae`	float	≥ 0.0
`dSASA`	float	≥ 0.0
`interface_sasa_pct`	float	0.0 to 100.0
`interface_hydrophobicity`	float	0.0 to 100.0
`surface_hydrophobicity`	float	0.0 to 1.0
`shape_complementarity`	float	0.0 to 1.0
`n_interface_residues`	float	≥ 0.0
`binder_helix_pct`	float	0.0 to 100.0
`binder_betasheet_pct`	float	0.0 to 100.0
`binder_loop_pct`	float	0.0 to 100.0
`interface_helix_pct`	float	0.0 to 100.0
`interface_betasheet_pct`	float	0.0 to 100.0
`interface_loop_pct`	float	0.0 to 100.0
`hotspot_rmsd`	float	≥ 0.0
`target_rmsd`	float	≥ 0.0
`binder_rmsd`	float	≥ 0.0
`unrelaxed_clashes`	float	≥ 0.0
`relaxed_clashes`	float	≥ 0.0

Applications

This tool is appropriate for de novo binder generation against a structurally characterised target where no curated antibody scaffold or pre-existing binder is available, and where a permissive license is required for hosting or commercial use. Representative applications include designing miniprotein binders against cell-surface receptors, generating binders that occlude a specific epitope or active site through hotspot targeting, and producing structurally diverse binder candidates for downstream therapeutic engineering.

Usage Tips

Provide a hotspot residue list when targeting a defined epitope. Set target_hotspot_residues to a comma-separated list of 1-indexed residue positions on the target structure, with ranges supported (for example "1-10,56,78"). Without hotspots the binder may land anywhere on the target surface.
binder_lengths defaults to (65, 150) residues, matching the upstream default. Binders below approximately 50 residues are effectively peptides and the AlphaFold2 multimer signal weakens; binders above approximately 200 residues introduce significant GPU memory and runtime costs.
weights_helicity controls the helix bias during hallucination. The default of -0.3 is a mild anti-helix bias because AlphaFold2 tends to over-produce alpha-helical bundles. Set a positive value to encourage helices, or set random_helicity=True to randomise the sign per trajectory.
optimise_beta=True (the default) adds extra hallucination iterations and AlphaFold2 recycles when a trajectory looks beta-heavy. Keep this enabled for any target that may favour beta-strand interfaces, such as immunoglobulin folds.
filter_overrides lets you relax or tighten individual filter thresholds. Pass a dict keyed by upstream metric name (such as "Average_i_pTM") and valued as a filter dict ({"threshold": 0.45, "higher": True}). Lower the interface pTM or shape complementarity threshold first if zero designs are accepted on a hard target. Note that filters on PyRosetta-only metrics (Average_dG, H-bond counts, Average_PackStat) pass trivially against placeholder values, so design selection is driven by the AlphaFold2 confidence, shape-complementarity, SASA, and clash metrics.
Production runs use number_of_final_designs=100 and max_trajectories=False. For a smoke test, set both to 1 together with reduced iteration counts (for example soft_iterations=10, temporary_iterations=5, hard_iterations=2, greedy_iterations=2).
The output is iterable. Iterating directly over the returned FreeBindCraftOutput yields each accepted FreeBindCraftDesign in turn, and len(result) returns the number of accepted designs.
Complementary tools cover adjacent design tasks. Reach for proteinmpnn-sample when an existing target-bound binder backbone only needs sequence redesign, rfdiffusion3-design when only a backbone is required without an accompanying sequence, and bindcraft-design (the PyRosetta variant) when Rosetta interface energetics are required for academic, non-commercial use.

Toolkit Notes

These apply to the FreeBindCraft tool in this toolkit (freebindcraft-design).

The pipeline runs on a single GPU per trajectory and benefits from 32 to 80 GB of GPU memory. AlphaFold2 multimer dominates the memory footprint and scales with the combined target plus binder length. To parallelise across multiple GPUs, run multiple instances of freebindcraft-design concurrently through a ToolPool.
OpenMM relaxation is GPU-accelerated and runs noticeably faster than PyRosetta’s CPU-bound FastRelax, so a FreeBindCraft trajectory typically completes faster than the equivalent BindCraft trajectory.
The first run downloads approximately 5.5 GB of AlphaFold2 weights together with the ColabDesign and FreeBindCraft repositories. Subsequent runs reuse the cached weights, which are shared with the proto-tools alphafold2 toolkit.

Example notebook: See the full working example for a copy-paste-ready walkthrough.

Infrastructure Guides

The following guides cover how to run tools efficiently and at scale.

Tool Persistence

Keep a tool’s model warm across calls instead of reloading it every invocation.

Device Management

How GPUs are allocated to tools and how to target specific devices.

Parallel Execution

Fan a batch of inputs out across multiple GPUs.

​Background

​PyRosetta-free metrics

​Learning Resources

​Tools

​FreeBindCraft Binder Design (freebindcraft-design)

​API Reference

​Applications

​Usage Tips

​Toolkit Notes

​Infrastructure Guides

Tool Persistence

Device Management

Parallel Execution

Background

PyRosetta-free metrics

Learning Resources

Tools

FreeBindCraft Binder Design (`freebindcraft-design`)

API Reference

Applications

Usage Tips

Toolkit Notes

Infrastructure Guides