← All tools
Binder design · Protein Design · Antibody design

BoltzGen

(Work in Progress)

Designs proteins and peptides that bind to a wide range of biomolecular targets.

It unifies design and structure prediction, resulting in a single model that also achieves state-of-the-art folding performance. BoltzGen was developed at MIT and experimentally validated in a large-scale distributed effort involving multiple academic and industry labs. Explicitly focuses our experimental validation on targets that are highly dissimilar to any proteins for which bound structures exist.

Home page ↗ Paper ↗ Source code ↗ License ↗

No jobs before this; it will run immediately.

Ready
Uploaded structure file for the binding target.
Comma-separated chain IDs to include from the target file; blank includes all.
One "chain:residues" per line, e.g. "A:5..7,13"; BoltzGen residue ranges use "..".
One "chain:residues" per line, residues the binder must not contact.
Leave blank to design a free peptide with no target.
A fixed sequence, a length range like "10..20", or a mix using BoltzGen's pattern syntax.
JSON list, e.g. [{"position1": 2, "position2": 10}], 1-indexed into the peptide. Positions must land on a literal residue already written into the sequence pattern (e.g. a "C"), not inside a design-length run.
BoltzGen sequence pattern mixing fixed residues and design-length runs.
JSON list of 1-indexed {"position1", "position2"} pairs. Positions must land on a literal residue already written into the sequence pattern (e.g. one of its "C"s), not inside a design-length run.
A solved nanobody scaffold to graft new CDR loops onto.
Residue ranges in the framework file, BoltzGen syntax (e.g. "26..34,52..59,98..118").
Residue ranges to strip from the template before redesign, e.g. to shorten a loop.
JSON list, e.g. [{"position": 26, "lengths": "1..5"}], inserted at a framework residue.
Leave at 0 for BoltzGen's own default.
One-letter amino-acid codes to avoid when designing sequences.
Ready