{"id":"5b74c12a-2120-4e93-b93d-b3d881317ce0","shortId":"YV6akw","kind":"skill","title":"binder-design","tagline":"Guidance for choosing the right protein binder design tool. Use this skill when: (1) Deciding between BoltzGen, BindCraft, or RFdiffusion, (2) Planning a binder design campaign, (3) Understanding trade-offs between different approaches, (4) Selecting tools for specific target typ","description":"# Binder Design Tool Selection\n\n## Decision tree\n\n```\nDe novo binder design?\n│\n├─ Standard target → BoltzGen (recommended)\n│   All-atom output (no separate ProteinMPNN step needed)\n│   Better for ligand/small molecule binding\n│   Single-step design (backbone + sequence + side chains)\n│\n├─ Need diversity/exploration → RFdiffusion + ProteinMPNN\n│   Maximum backbone diversity\n│   Two-step: backbone then sequence\n│\n├─ Integrated validation → BindCraft\n│   Built-in AF2 validation\n│   End-to-end pipeline\n│\n├─ Ligand binding → BoltzGen ✓\n│   All-atom diffusion handles ligand context\n│\n├─ Peptide/nanobody → Germinal\n│   VHH/nanobody design\n│   Germline-aware optimization\n│\n└─ Antibody/Nanobody\n    +-- VHH design --> germinal skill\n```\n\n## Tool comparison\n\n| Tool | Strengths | Weaknesses | Best For |\n|------|-----------|------------|----------|\n| BoltzGen | All-atom, single-step, ligand-aware | Higher GPU requirement | Standard (recommended) |\n| BindCraft | End-to-end, built-in AF2 validation | Less diverse | Production campaigns |\n| RFdiffusion | High diversity, fast | Requires ProteinMPNN | Exploration, diversity |\n| Germinal | Nanobody/VHH design | Specialized | Antibody optimization |\n\n## Recommended Pipeline: BoltzGen → Chai → QC\n\nBoltzGen provides all-atom design with built-in side-chain packing:\n\n```\nTarget → BoltzGen → Validate → Filter\n (pdb)  (all-atom)   (chai)     (qc)\n```\n\n### 1. Target preparation\n```bash\n# Fetch structure from PDB\n# Use pdb skill for guidance\n```\n- Trim to binding region + 10A buffer\n- Remove waters and ligands\n- Renumber chains if needed\n\n### 2. Hotspot selection\n- Choose 3-6 exposed residues\n- Prefer charged/aromatic residues\n- Cluster spatially (within 10-15A)\n\n### 3. Design with BoltzGen (Recommended)\n\nFirst, create a YAML config file (e.g., `binder.yaml`):\n```yaml\nentities:\n  - protein:\n      id: B\n      sequence: 70..100\n\n  - file:\n      path: target.cif\n      include:\n        - chain:\n            id: A\n      binding_types:\n        - chain:\n            id: A\n            binding: 45,67,89\n```\n\nThen run:\n```bash\nmodal run modal_boltzgen.py \\\n  --input-yaml binder.yaml \\\n  --protocol protein-anything \\\n  --num-designs 50\n```\n\n**Why BoltzGen?**\n- All-atom output (no separate ProteinMPNN step needed)\n- Better for ligand/small molecule binding\n- Single-step design (backbone + sequence + side chains)\n\n### 4. Alternative: RFdiffusion Pipeline\nFor maximum diversity or when backbone-only is preferred:\n```bash\n# Step 1: Backbone generation\nmodal run modal_rfdiffusion.py \\\n  --pdb target.pdb \\\n  --contigs \"A1-150/0 70-100\" \\\n  --hotspot \"A45,A67,A89\" \\\n  --num-designs 500\n\n# Step 2: Sequence design\nmodal run modal_ligandmpnn.py \\\n  --pdb-path backbone.pdb \\\n  --num-seq-per-target 16 \\\n  --sampling-temp 0.1\n```\n\n### 5. Validation\n```bash\nmodal run modal_chai1.py \\\n  --input-faa sequences.fasta \\\n  --out-dir predictions/\n```\n\n### 6. Filtering\nApply standard thresholds:\n- pLDDT > 0.80\n- ipTM > 0.50\n- PAE_interface < 10\n- scRMSD < 2.0 A\n\nSee protein-qc skill for details.\n\n## Number of designs\n\n| Stage | Count | Purpose |\n|-------|-------|---------|\n| Backbone generation | 500-1000 | Diversity |\n| Sequences per backbone | 8-16 | Sequence space |\n| AF2 predictions | All | Validation |\n| After filtering | 50-200 | Candidates |\n| Experimental testing | 10-50 | Final selection |\n\n## Common mistakes\n\n### Wrong hotspots\n- Using buried residues\n- Too many hotspots (over-constrain)\n- Wrong chain/residue numbers\n\n### Insufficient diversity\n- Too few designs generated\n- Low temperature in ProteinMPNN\n- Not exploring multiple backbones\n\n### Poor target preparation\n- Including full protein instead of binding region\n- Missing important structural features\n- Wrong protonation states\n\n## Timeline guide\n\n| Step | Compute Time |\n|------|--------------|\n| RFdiffusion (500 designs) | 2-4 hours |\n| ProteinMPNN (8000 sequences) | 1-2 hours |\n| AF2 prediction (8000 sequences) | 12-24 hours |\n| Filtering and analysis | 1-2 hours |\n\nTotal: 1-2 days of compute","tags":["binder","design","protein","skills","adaptyvbio","agent-skills","claude-code","protein-design","protein-engineering"],"capabilities":["skill","source-adaptyvbio","skill-binder-design","topic-agent-skills","topic-claude-code","topic-protein-design","topic-protein-engineering"],"categories":["protein-design-skills"],"synonyms":[],"warnings":[],"endpointUrl":"https://skills.sh/adaptyvbio/protein-design-skills/binder-design","protocol":"skill","transport":"skills-sh","auth":{"type":"none","details":{"cli":"npx skills add adaptyvbio/protein-design-skills","source_repo":"https://github.com/adaptyvbio/protein-design-skills","install_from":"skills.sh"}},"qualityScore":"0.513","qualityRationale":"deterministic score 0.51 from registry signals: · indexed on github topic:agent-skills · 126 github stars · SKILL.md body (4,021 chars)","verified":false,"liveness":"unknown","lastLivenessCheck":null,"agentReviews":{"count":0,"score_avg":null,"cost_usd_avg":null,"success_rate":null,"latency_p50_ms":null,"narrative_summary":null,"summary_updated_at":null},"enrichmentModel":"deterministic:skill-github:v1","enrichmentVersion":1,"enrichedAt":"2026-05-02T12:54:48.114Z","embedding":null,"createdAt":"2026-04-18T22:10:02.303Z","updatedAt":"2026-05-02T12:54:48.114Z","lastSeenAt":"2026-05-02T12:54:48.114Z","tsv":"'-100':361 '-1000':436 '-15':251 '-150':358 '-16':442 '-2':522,535,539 '-200':452 '-24':529 '-4':516 '-50':457 '-6':241 '/0':359 '0.1':390 '0.50':413 '0.80':411 '1':17,209,348,521,534,538 '10':250,416,456 '100':273 '10a':226 '12':528 '16':386 '2':24,236,371,515 '2.0':418 '3':30,240,253 '4':38,332 '45':287 '5':391 '50':307,451 '500':369,435,513 '6':405 '67':288 '70':272,360 '8':441 '8000':519,526 '89':289 'a1':357 'a45':363 'a67':364 'a89':365 'af2':100,160,445,524 'all-atom':59,110,138,187,204,310 'altern':333 'analysi':533 'antibodi':178 'antibody/nanobody':125 'anyth':303 'appli':407 'approach':37 'atom':61,112,140,189,206,312 'awar':123,146 'b':270 'backbon':77,86,91,328,342,349,433,440,489 'backbone-on':341 'backbone.pdb':380 'bash':212,292,346,393 'best':135 'better':68,319 'bind':72,108,224,281,286,323,498 'bindcraft':21,96,152 'binder':2,10,27,45,53 'binder-design':1 'binder.yaml':265,299 'boltzgen':20,57,109,137,182,185,200,256,309 'buffer':227 'built':98,158,193 'built-in':97,157,192 'buri':465 'campaign':29,165 'candid':453 'chai':183,207 'chain':80,197,233,278,283,331 'chain/residue':474 'charged/aromatic':245 'choos':6,239 'cluster':247 'common':460 'comparison':131 'comput':510,542 'config':262 'constrain':472 'context':116 'contig':356 'count':431 'creat':259 'day':540 'de':51 'decid':18 'decis':49 'design':3,11,28,46,54,76,120,127,176,190,254,306,327,368,373,429,480,514 'detail':426 'differ':36 'diffus':113 'dir':403 'divers':87,163,168,173,338,437,477 'diversity/exploration':82 'e.g':264 'end':103,105,154,156 'end-to-end':102,153 'entiti':267 'experiment':454 'explor':172,487 'expos':242 'faa':399 'fast':169 'featur':503 'fetch':213 'file':263,274 'filter':202,406,450,531 'final':458 'first':258 'full':494 'generat':350,434,481 'germin':118,128,174 'germlin':122 'germline-awar':121 'gpu':148 'guid':508 'guidanc':4,221 'handl':114 'high':167 'higher':147 'hotspot':237,362,463,469 'hour':517,523,530,536 'id':269,279,284 'import':501 'includ':277,493 'input':297,398 'input-faa':397 'input-yaml':296 'instead':496 'insuffici':476 'integr':94 'interfac':415 'iptm':412 'less':162 'ligand':107,115,145,231 'ligand-awar':144 'ligand/small':70,321 'low':482 'mani':468 'maximum':85,337 'miss':500 'mistak':461 'modal':293,351,374,394 'modal_boltzgen.py':295 'modal_chai1.py':396 'modal_ligandmpnn.py':376 'modal_rfdiffusion.py':353 'molecul':71,322 'multipl':488 'nanobody/vhh':175 'need':67,81,235,318 'novo':52 'num':305,367,382 'num-design':304,366 'num-seq-per-target':381 'number':427,475 'off':34 'optim':124,179 'out-dir':401 'output':62,313 'over-constrain':470 'pack':198 'pae':414 'path':275,379 'pdb':203,216,218,354,378 'pdb-path':377 'peptide/nanobody':117 'per':384,439 'pipelin':106,181,335 'plan':25 'plddt':410 'poor':490 'predict':404,446,525 'prefer':244,345 'prepar':211,492 'product':164 'protein':9,268,302,422,495 'protein-anyth':301 'protein-qc':421 'proteinmpnn':65,84,171,316,485,518 'protocol':300 'proton':505 'provid':186 'purpos':432 'qc':184,208,423 'recommend':58,151,180,257 'region':225,499 'remov':228 'renumb':232 'requir':149,170 'residu':243,246,466 'rfdiffus':23,83,166,334,512 'right':8 'run':291,294,352,375,395 'sampl':388 'sampling-temp':387 'scrmsd':417 'see':420 'select':39,48,238,459 'separ':64,315 'seq':383 'sequenc':78,93,271,329,372,438,443,520,527 'sequences.fasta':400 'side':79,196,330 'side-chain':195 'singl':74,142,325 'single-step':73,141,324 'skill':15,129,219,424 'skill-binder-design' 'source-adaptyvbio' 'space':444 'spatial':248 'special':177 'specif':42 'stage':430 'standard':55,150,408 'state':506 'step':66,75,90,143,317,326,347,370,509 'strength':133 'structur':214,502 'target':43,56,199,210,385,491 'target.cif':276 'target.pdb':355 'temp':389 'temperatur':483 'test':455 'threshold':409 'time':511 'timelin':507 'tool':12,40,47,130,132 'topic-agent-skills' 'topic-claude-code' 'topic-protein-design' 'topic-protein-engineering' 'total':537 'trade':33 'trade-off':32 'tree':50 'trim':222 'two':89 'two-step':88 'typ':44 'type':282 'understand':31 'use':13,217,464 'valid':95,101,161,201,392,448 'vhh':126 'vhh/nanobody':119 'water':229 'weak':134 'within':249 'wrong':462,473,504 'yaml':261,266,298","prices":[{"id":"60b11513-740e-43fe-af77-5192026992dc","listingId":"5b74c12a-2120-4e93-b93d-b3d881317ce0","amountUsd":"0","unit":"free","nativeCurrency":null,"nativeAmount":null,"chain":null,"payTo":null,"paymentMethod":"skill-free","isPrimary":true,"details":{"org":"adaptyvbio","category":"protein-design-skills","install_from":"skills.sh"},"createdAt":"2026-04-18T22:10:02.303Z"}],"sources":[{"listingId":"5b74c12a-2120-4e93-b93d-b3d881317ce0","source":"github","sourceId":"adaptyvbio/protein-design-skills/binder-design","sourceUrl":"https://github.com/adaptyvbio/protein-design-skills/tree/main/skills/binder-design","isPrimary":false,"firstSeenAt":"2026-04-18T22:10:02.303Z","lastSeenAt":"2026-05-02T12:54:48.114Z"}],"details":{"listingId":"5b74c12a-2120-4e93-b93d-b3d881317ce0","quickStartSnippet":null,"exampleRequest":null,"exampleResponse":null,"schema":null,"openapiUrl":null,"agentsTxtUrl":null,"citations":[],"useCases":[],"bestFor":[],"notFor":[],"kindDetails":{"org":"adaptyvbio","slug":"binder-design","github":{"repo":"adaptyvbio/protein-design-skills","stars":126,"topics":["agent-skills","claude-code","protein-design","protein-engineering"],"license":"mit","html_url":"https://github.com/adaptyvbio/protein-design-skills","pushed_at":"2026-01-19T13:06:29Z","description":"Claude Code skills for protein design","skill_md_sha":"09ae83c0ccb338876332d1c69f6260f38feee0f3","skill_md_path":"skills/binder-design/SKILL.md","default_branch":"main","skill_tree_url":"https://github.com/adaptyvbio/protein-design-skills/tree/main/skills/binder-design"},"layout":"multi","source":"github","category":"protein-design-skills","frontmatter":{"name":"binder-design","license":"MIT","description":"Guidance for choosing the right protein binder design tool. Use this skill when: (1) Deciding between BoltzGen, BindCraft, or RFdiffusion, (2) Planning a binder design campaign, (3) Understanding trade-offs between different approaches, (4) Selecting tools for specific target types.  For specific tool parameters, use the individual tool skills (boltzgen, bindcraft, rfdiffusion, etc.)."},"skills_sh_url":"https://skills.sh/adaptyvbio/protein-design-skills/binder-design"},"updatedAt":"2026-05-02T12:54:48.114Z"}}