RFdiffusion Online: Generate Peptide Binders to a Target Protein Without Coding
How to design candidate peptide binders against a target protein using RFdiffusion in ChemOrchestra — no local GPU, no coding.
Jul 18, 2026
Model
RFdiffusion
Input
Target protein structure
Output
Candidate peptide binders
What RFdiffusion does
RFdiffusion is a generative model for protein design — instead of finding a binder among existing molecules the way docking or virtual screening does, it generates new peptide structures directly, conditioned on a target protein's binding site. That makes it useful when no known binder exists yet, or when you want to explore binding modes outside what's already been characterized.
Running Binder Design in ChemOrchestra
Provide a target protein structure to the Binder Design node and run it to generate a set of candidate peptide binders against that target. The generated structures render directly in the workflow canvas, ready for further evaluation.
Evaluating candidate binders before synthesis
A generated binder is a hypothesis, not a validated design. Fold-check candidates with ESMfold or evaluate the complex with Boltz-2 co-folding to assess whether the predicted binding geometry is plausible, and run ADMET prediction to check developability, before committing any candidate to synthesis.
When to design a binder vs. screen for a small molecule
Binder design and small-molecule approaches solve the same problem — engaging a target — through different modalities. Peptide binders can access flatter, more extended binding surfaces that are difficult for small molecules to engage, at the cost of the oral bioavailability small molecules typically have. Use Docking or Virtual Screening when a small-molecule approach fits the target; reach for Binder Design when the interface calls for something a small molecule can't cover.