How to Use BoltzGen

Generate high-affinity binders for proteins, nucleic acids, and small molecules using an all-atom diffusion model.

Key run controls include Mode, Minimum Length, Maximum Length, and Number Designs, making protocol choices explicit and reproducible for collaborative projects.

The 1.0 results page exposes analysis sections like Final Designs and All Designs, output artifacts such as final.csv and all.csv, and interactive views including protein viewer molstar and final structure select for rapid construct prioritization and export-ready reporting.

BoltzGen is a universal all-atom generative model for biomolecular binder design. It unifies sequence design and structure prediction within a single diffusion-based framework, enabling the creation of high-affinity binders against proteins, nucleic acids, and small molecules. Through a geometry-based representation and a flexible design specification language, BoltzGen allows precise control over covalent bonds, structural motifs, and binding interfaces while maintaining state-of-the-art folding accuracy.

Core capabilities include Unifies structure prediction and binder design in a single all-atom diffusion framework., Supports protein, peptide, nanobody, and cyclic peptide binders targeting proteins, nucleic acids, or small molecules., and Enables controllable generation of diverse and designable binders through architecture-level conditioning and tunable noise schedules., which directly shape how outputs should be interpreted for this method.

In practice, interpretation proceeds through inputs (Target Structure, Target Chains, and Binding Sites), result sections (Final Designs and All Designs), files (final.csv and all.csv), and visual components (protein viewer molstar and final structure select), which makes BoltzGen outputs easier to triage and act on across large job batches.

• Study-fit inputs: BoltzGen accepts Target Structure, Target Chains, Binding Sites, and Non-Binding Sites, reducing preprocessing friction and preserving experimental context.
• Protocol control: Researchers can tune Mode, Minimum Length, Maximum Length, and Number Designs to match assay constraints, confidence thresholds, and downstream validation plans.
• Readable evidence: Results are presented through Final Designs and All Designs, final.csv and all.csv, and protein viewer molstar and final structure select, improving cross-run comparison and scientific communication.
• Faster iteration: Managed execution on Neurosnap removes infrastructure overhead so teams can focus on protein engineering and variant selection rather than deployment and environment maintenance.