Design antibody, nanobody, and de novo binder backbones conditioned on target structures.
npx skills add https://github.com/001TMF/blatant-why --skill boltzgen-001tmf
Please help me install this Agent Skill. Skill: boltzgen Source: https://github.com/001TMF/blatant-why/tree/main/templates/.claude/skills/boltzgen Command: npx skills add https://github.com/001TMF/blatant-why --skill boltzgen-001tmf
BoltzGen streamlines binder design by generating antibody, nanobody, and de novo binder backbones and sequences conditioned on a target structure and a defined epitope, reducing the need for iterative experimental design.
Provide a target structure and epitope; BoltzGen will generate designed backbones and sequences for downstream screening.
High-intent search queries and answers about installing and using this skill.
BoltzGen generates antibody backbones and sequences conditioned on a target structure and defined binding residues. You provide a target CIF/PDB and an entities YAML spec with epitope ranges to produce refolded designs using the boltzgen CLI.
Yes, BoltzGen uses all-atom diffusion backbone design to support single-domain VHH nanobodies, full antibody Fab, and de novo miniproteins. Sequences are assigned via AntiFold, followed by Protenix refolding to produce scoring metrics.
Designed binders are scored using ipTM, pTM, pLDDT, ipSAE, and CA-RMSD metrics. Protenix refolding generates these metrics to help filter and rank candidates during downstream screening.
BoltzGen supports flexible deployment across local GPUs requiring at least 24 GB, HPC RunPod, or Tamarind cloud. This allows you to run binder design workflows in your preferred computing environment.
You need a target structure in CIF or PDB format and an entities YAML spec defining epitope ranges. These inputs allow the boltzgen CLI to condition the diffusion design on your specific target binding residues.
AntiFold performs inverse folding to assign sequences to the designed backbones generated by BoltzGen. Protenix then refolds these sequences to evaluate structural quality using ipTM, pTM, pLDDT, ipSAE, and RMSD metrics.
junior1pDesign antigen-conditioned antibody and nanobody CDR loops from X-masked FASTA inputs.
alex-hhCross-validate 15-PGDH binder designs using Boltz-2 structure prediction.
PKU-YuanGroupConverts PDF documents to clean, structured Markdown.
junior1pGuide protein binder design workflows with BoltzGen, BindCraft, and Chai validation.
