How to Use BindCraft2

Commercially Available Online Web Server

BindCraft2 guide to protein binder campaigns, VHH and peptide formats, epitope targeting, campaign limits, and ranked design interpretation.

BindCraft2, or BindCraft 2.0, is a campaign-based protein binder design workflow combining AlphaFold2-guided optimization, ProteinMPNN sequence redesign, held-out AlphaFold validation, and structural filtering. On Neurosnap it supports several binder architectures, including miniproteins, peptides, cyclic peptides, oligomeric and multidomain binders, VHHs, ankyrin repeat proteins, scFvs, and Fabs.

A campaign searches through candidate trajectories and attempts to return a requested number of accepted designs. That count is a goal, not a guarantee: challenging targets, restrictive objectives, or stopping conditions can yield partial results or no accepted candidates.

When to choose BindCraft2

Use it when the experiment calls for a specific binder format, epitope, or campaign-level validation workflow. Optional properties such as focused epitope targeting, termini placement, and disulfide stapling can express design objectives, but each adds assumptions that should match the planned experiment.

For a simpler backbone-first exploration, compare RFdiffusion. For the separate earlier implementation and its controls, see BindCraft. BindCraft2 can consume substantial compute; select it for a defined design problem and review campaign limits before submission. Computational acceptance does not establish experimental binding, safety, or developability.

How BindCraft2 Works

Set the target, hotspots, and coldspots

Upload Target Structure and choose Target Chains, leaving the selection blank only when every uploaded chain should be included. Inspect the biological assembly and identify a surface that the intended binder format can physically reach.

Use Hotspots to encourage contact with selected residues and Coldspots to discourage contact with others. Selections use the uploaded PDB numbering, for example A56,A60-70. Keep a residue map if you crop or renumber the target. Focused Epitope concentrates design around supplied hotspots; it should represent a real epitope objective rather than an arbitrary extra restriction.

Cropping a large target may reduce the computational problem, but it can also expose artificial surfaces. Check shortlisted complexes against the complete target, including neighboring domains and oligomer partners. The input removes nucleic acids, heteroatoms, and waters, so assess omitted glycans and cofactors separately when considering accessibility and specificity.

Choose the binder format and compatible length

Binder Format Design context
binder / large_binder De novo protein binders with different size regimes.
peptide / cyclic_peptide Peptide-oriented designs; cyclization must also be feasible in the intended synthesis and assay.
homo_oligomer / multidomain Architectures requiring explicit assembly or domain organization.
VHH / ARP Scaffold-based nanobody or ankyrin-repeat formats.
scFv / Fab Antibody-fragment scaffold presets with format-specific context.

Set Minimum Binder Length and Maximum Binder Length where the format samples length. Scaffold-based formats impose additional architectural assumptions, so an arbitrary length range is not interchangeable across all modalities. Oligomer Copies is relevant to homo-oligomer campaigns; leave it at 1 for ordinary single-chain binders.

Conformational Mode offers induced-fit and fold-switching objectives. These introduce a more specific structural hypothesis and should be enabled only when that behavior is part of the design question. A predicted conformational change is not experimental evidence that the protein switches states.

Add biological objectives selectively

Humanize Binder, Protease Stable, and Disulfide Staple steer sequence or structural objectives when supported by the selected modality. They do not demonstrate low immunogenicity, measured protease resistance, or successful disulfide formation.

Mixed Topology biases secondary-structure balance. Nearby Termini and Accessible Termini can help express construct requirements, such as fusion compatibility. Verify the actual resulting geometry against the intended fusion partner or presentation system instead of assuming a preset guarantees compatibility.

Avoid enabling every objective simultaneously. When a campaign struggles, a smaller set of scientifically necessary constraints makes it easier to determine whether the difficulty comes from the target, architecture, or acceptance criteria.

Plan campaign size and stopping conditions

Number of Final Designs counts accepted candidates, not the number of attempted trajectories. Maximum Trajectories caps design attempts, while Accepted Design Timeout (Hours) limits time without a newly accepted design. Each accepted design resets that timeout; it is not a fixed total-runtime cap. Additional service stopping conditions also apply.

Record Campaign Seed, target preparation, selected format, objectives, and stopping settings. A seed helps document the campaign but does not guarantee identical numerical behavior across all hardware and software environments. Increasing the requested design count or allowing longer exploration can increase compute consumption without ensuring success.

Understand trajectories, refolded candidates, and ranked designs

The results separate stages of the search. Trajectories describe design attempts, refolded candidates capture redesigned sequences and filtering outcomes, and ranked results organize accepted designs. Inspect the ranked-design summary alongside the complex structures; a ranking alone cannot explain whether a binder contacts the intended epitope.

Held-out AlphaFold validation adds a check beyond the model used during optimization, but the assessment remains computational and can share model biases. Compare the binder fold, target conformation, interface geometry, and candidate diversity. Retain reasons for rejection when diagnosing why a campaign yielded few accepted designs.

Example: design a VHH to a specified epitope

Prepare the relevant target assembly, select the intended chain, and map the epitope to uploaded residue numbers. Choose VHH, add hotspots, and use only the objectives required by the intended assay. Set a deliberate accepted-design goal and trajectory limit, then inspect both accepted candidates and failure patterns.

Shortlist structurally and sequentially diverse candidates. Review any exposed hydrophobic patches and scaffold changes before expression, then test binding and specificity experimentally. The design format alone does not establish therapeutic suitability.

BindCraft2 questions and troubleshooting

Why are there fewer final designs than requested? The requested count can be unmet when limits are reached or candidates fail acceptance criteria. Review partial output and failure trends before expanding the campaign.

Should I increase the timeout first? Check target preparation, numbering, accessible surface, format compatibility, and restrictive objectives before spending more on the same setup.

Are humanized designs experimentally validated? No. Humanization is a design objective; immunogenicity and other biological properties require appropriate independent assessment.

What is Neurosnap?

Neurosnap is the leading platform for bioinformatics and computational science focused on expanding access to powerful modeling and simulation tools. Because many state-of-the-art machine learning systems remain complex to install, configure, and scale, Neurosnap offers a clean, browser-based workspace that removes the burden of infrastructure management, dependency conflicts, and command-line tooling.

Built for biologists, chemists, and cross-disciplinary scientists, the platform enables advanced computational workflows without requiring expertise in software engineering or cloud architecture. Researchers can launch analyses through an intuitive interface, connect programmatically through a comprehensive API, and rely on automated resource management to scale workloads efficiently. By taking care of the underlying compute and operational complexity, Neurosnap allows teams to devote their energy to scientific progress and faster iteration. Security and data protection remain foundational principles, with clear safeguards outlined in our Terms of Use and Privacy Policy to ensure your work stays protected.

Advancing Discovery with BindCraft2 on Neurosnap

Using BindCraft2 on Neurosnap could drastically accelerate multimodal protein binder design with campaign-level ranking, refolding, and structural filtering.

  • Broad modality coverage: BindCraft2 handles de novo binders, peptides, cyclic peptides, oligomers, scaffolded binders, and antibody-like formats in one campaign model.
  • Epitope-aware inputs: Target chains, hotspots, coldspots, focused epitope targeting, and termini objectives let researchers align the search with biological constraints.
  • Integrated validation: ProteinMPNN redesign, held-out structure prediction, and structural filters help prioritize candidates beyond raw sequence generation.
  • Readable campaign outputs: Ranked designs, refolded candidates, trajectory summaries, and predicted structures make it easier to triage binders for expression, biophysical screening, and downstream optimization.

How to Use BindCraft2 on Neurosnap

To harness the capabilities of BindCraft2, researchers can follow this streamlined workflow within Neurosnap:

  1. Access Neurosnap: Start by logging in to the Neurosnap website.
  2. Select Tool: From the list of available tools, choose BindCraft2.
  3. Provide Inputs: Provide all the inputs specified within the submission panel and optionally configure the tool as desired.
  4. Run Tool: Submit the BindCraft2 job and Neurosnap will execute it in the cloud, automatically notifying you as soon as your results are ready.
  5. Review Output: Explore your results through rich visualizations, including figures, plots, and interactive views designed to help you analyze findings with clarity and confidence.

Citations

Please cite the original work when using BindCraft2 in publications or research outputs.

PacesaLab. BindCraft2: Easy to use suite for designing a wide range of protein binder modalities. https://github.com/PacesaLab/BindCraft2

Pacesa, M., Nickel, L., Schellhaas, C. et al. One-shot design of functional protein binders with BindCraft. Nature (2025). https://doi.org/10.1038/s41586-025-09429-6

Jumper et al. "Highly accurate protein structure prediction with AlphaFold." Nature (2021) doi: 10.1038/s41586-021-03819-2

Dauparas, J., et al. "Robust deep learning based protein sequence design using ProteinMPNN." Science (2022) doi: 10.1126/science.add2187

Neurosnap Inc. (2022). Neurosnap: An online platform for computational biology and chemistry. Available at: https://neurosnap.ai/

Similar Services

Explore related tools that support similar research workflows:


Proudly supporting 50,000+ scientists worldwide, including 7,000+ leading biotech and global biopharma organizations.

Making Scientific Research
Faster & Easier

Register for free — upgrade anytime.

Interested in getting a license? Contact Sales.

Try Free