Use BindCraft Online
Commercially Available BindCraft No-Code Web Server
BindCraft: Open-Source De Novo Protein Binder Design with AlphaFold2
BindCraft is an open-source, automated pipeline for de novo protein binder design that uses AlphaFold2 weights to produce binders, often at nanomolar affinity, without high-throughput screening. It was developed by Martin Pacesa, Bruno E. Correia and colleagues at EPFL, the University of Zurich, MIT and other institutions, and described in the 2024 bioRxiv preprint "BindCraft: one-shot design of functional protein binders". The code is released under the MIT license.
Designing a protein that binds a chosen target has traditionally meant screening large libraries and optimizing experimentally. BindCraft aims for a "one design, one binder" approach, with reported experimental success rates of 10 to 100% across targets.
Property | Detail |
|---|---|
Core engine | AF2-multimer backpropagation (ColabDesign), ProteinMPNN, PyRosetta scoring |
Inputs | Target structure, binder length range, number of designs, optional hotspots |
Reported success | 10 targets, 24.5-100% per target, 49.5% average |
Binder size | 60 to 240 residues discussed |
Compute | About 30 GB GPU memory for a 500-residue complex |
How BindCraft Designs Protein Binders
Hallucinate: backpropagation through AF2-multimer generates binder sequences and structures while co-folding the target.
Optimize sequences: MPNN_sol redesigns the binder core and surface while keeping the interface fixed.
Re-predict: designs are checked with the AF2 monomer model in single-sequence mode.
Filter: AF2 confidence metrics and Rosetta physics-based scores select designs.
Experimental Results: Binder Hit Rates and Affinities
Target | Designs tested to binders | Best affinity |
|---|---|---|
PD-1 | 53 to 13 | Apparent Kd below 1 nM |
PD-L1 | 9 to 7 | 615 nM |
CD45 | 16 to 4 | 14.7 nM |
Der f7 | 10 to 4 | 12.8 nM |
Bet v1 | 7 to 2 | 120 nM |
Structural and Functional Validation
Crystal structures were solved for binders to BBF-14, Der f7 and Der f21, and cryo-EM structures for SpCas9 binders. Designed binders reduced SpCas9 editing in HEK293T cells and reduced IgE binding to birch allergen in patient-derived samples.
What is Tamarind Bio?
Tamarind Bio is a no-code bioinformatics platform built to give life scientists and researchers access to powerful computational tools. Many cutting-edge machine learning models are hard to deploy and use. Tamarind provides an intuitive, web-based environment that removes the complexity of high-performance computing, software dependencies and command-line interfaces.
The platform is designed for biologists, chemists and other researchers who may not have a background in programming or cloud infrastructure but want to run models on their own data. Key features include:
A user-friendly graphical interface for setting up and launching experiments
A robust API for integration into existing research pipelines
An automated system for managing and scaling computational resources
Tamarind treats information and data security as a top priority, as detailed in its Trust Center and Terms of Service.
Accelerating Discovery with BindCraft on Tamarind Bio
Therapeutic and cell-surface targets: design binders against receptors such as PD-1, PD-L1 and CD45.
Allergen blocking: design binders to allergens such as Bet v1.
Genome-editing control: design binders that modulate SpCas9.
Unknown binding sites: let AF2 choose the site when no hotspot is given.
How to Use BindCraft on Tamarind Bio
Log in and open the tool: sign in at tamarind.bio and select BindCraft.
Provide the target: upload a target PDB, trimmed to the region of interest.
Choose chains and hotspots: specify target chains and optional hotspot residues, if exposed.
Set binder length: give the binder size range.
Set design count: the README recommends at least 100 designs passing all filters.
Run: submit the job.
Download outputs: collect relaxed structures, MPNN sequences and statistics.
Select candidates: the README suggests ordering the top 5 to 20 designs for testing.
Things to Keep in Mind
Large targets need high GPU memory and may need trimming.
Single-sequence AF2 filtering may discard some high-affinity binders.
The ipTM ranking metric predicts binding but not affinity.
Most top designs are alpha-helical; beta-sheet designs have lower in silico success.
Immunogenicity and delivery are open concerns for therapeutic use.
Source: Pacesa M, Nickel L, Schmidt J, et al., Correia BE, "BindCraft: one-shot design of functional protein binders", bioRxiv preprint (2024), DOI 10.1101/2024.09.30.615802. Code: https://github.com/martinpacesa/BindCraft (MIT license).