← All tools
Protein Design · Binder design

RFdiffusion

Generate protein backbones unconditionally, around a motif, or against a target.

Runs the official RFdiffusion inference script. Custom Contigs exposes the raw contig map directly; the other tasks build one for you from a target/binder/motif description.

Source code ↗
Input

Configure a small job

POST
RFdiffusion contig string without the brackets. For example 150-150 for an unconditional monomer, 5-15/A10-25/30-40 to scaffold a motif, or B1-100/0 100-100 for a binder.
Required by any contig that references chains of a structure.
Comma- or space-separated residue numbers on hotspot_chain, e.g. 100,101,102.
Comma-separated chain IDs from the target PDB to keep as context.
e.g. 70-100.
One "chain: residues" per line, e.g. "A: 20 21 23".
Comma- or space-separated residue numbers.
Comma-separated ranges on binder_chain to regenerate; everything else on it stays fixed.
One length range per designed_residues range, e.g. 5-9,3-8,10-25. Blank keeps each span's original length.
Comma-separated ranges of the motif to preserve.
One range per gap (one more than the number of motif ranges), e.g. 10-40,10-40,10-40. Blank defaults every gap to 10-40.
Comma-separated ranges on designed_chain to noise; other residues (and other chains) stay fixed.
Comma-separated ranges, in designed_chain's own residue numbering, to hold sequence identity constant even while noised.
A structure whose secondary-structure/block-adjacency pattern the new binder should follow.
e.g. C2, C4, C6, D2, tetrahedral, octahedral, icosahedral.
Divided evenly across the symmetric copies by RFdiffusion.
Comma-separated per-copy residue ranges, one per symmetric unit.
Raw potential spec, e.g. "type:substrate_contacts,s:1,r_0:8,rep_r_0:5.0,rep_s:2,rep_r_min:1".
Only used with a substrate_contacts potential, e.g. LLK.
Ready
Response

JSON result

{
  "status": "Ready for a request"
}