{"id":"4ae38127-c840-44be-92e1-3255b9691cdb","shortId":"5YgEYk","kind":"skill","title":"alphafold","tagline":"Validate protein designs using AlphaFold2 structure prediction. Use this skill when: (1) Validating designed sequences fold correctly, (2) Predicting binder-target complex structures, (3) Calculating confidence metrics (pLDDT, pTM, ipTM), (4) Self-consistency validation of design","description":"# AlphaFold2 Structure Validation\n\n## Prerequisites\n\n| Requirement | Minimum | Recommended |\n|-------------|---------|-------------|\n| Python | 3.8+ | 3.10 |\n| CUDA | 11.0+ | 12.0+ |\n| GPU VRAM | 32GB | 40GB (A100) |\n| RAM | 32GB | 64GB |\n| Disk | 100GB | 500GB (for databases) |\n\n## How to run\n\n> **First time?** See [Installation Guide](../../docs/installation.md) to set up Modal and biomodals.\n\n### Option 1: ColabFold (recommended for multimer)\n```bash\ncd biomodals\nmodal run modal_colabfold.py \\\n  --input-faa sequences.fasta \\\n  --out-dir output/\n```\n\n**GPU**: A100 (40GB) | **Timeout**: 3600s default\n\n### Option 2: Local installation\n```bash\ngit clone https://github.com/deepmind/alphafold.git\ncd alphafold\n\npython run_alphafold.py \\\n  --fasta_paths=query.fasta \\\n  --output_dir=output/ \\\n  --model_preset=monomer \\\n  --max_template_date=2026-01-01\n```\n\n### Option 3: ESMFold (fast single-chain)\n```bash\nmodal run modal_esmfold.py \\\n  --sequence \"MKTAYIAKQRQISFVK...\"\n```\n\n## Key parameters\n\n| Parameter | Default | Options | Description |\n|-----------|---------|---------|-------------|\n| `--model_preset` | monomer | monomer/multimer | Model type |\n| `--num_recycle` | 3 | 1-20 | Recycling iterations |\n| `--max_template_date` | - | YYYY-MM-DD | Template cutoff |\n| `--use_templates` | True | True/False | Use template search |\n\n## Output format\n\n```\noutput/\n├── ranked_0.pdb           # Best model\n├── ranked_1.pdb           # Second best\n├── ranking_debug.json     # Confidence scores\n├── result_model_1.pkl     # Full results\n├── msas/                  # MSA files\n└── features.pkl           # Input features\n```\n\n### Extracting metrics\n```python\nimport pickle\n\nwith open('result_model_1.pkl', 'rb') as f:\n    result = pickle.load(f)\n\nplddt = result['plddt']\nptm = result['ptm']\niptm = result.get('iptm', None)  # Multimer only\npae = result['predicted_aligned_error']\n```\n\n## Sample output\n\n### Successful run\n```\n$ python run_alphafold.py --fasta_paths complex.fasta --model_preset multimer\n[INFO] Running MSA search...\n[INFO] Running model 1/5...\n[INFO] Running model 5/5...\n[INFO] Relaxing structures...\n\nResults:\n  ranked_0.pdb:\n    pLDDT: 87.3 (mean)\n    pTM: 0.78\n    ipTM: 0.62\n    PAE (interface): 8.5\n\nSaved to output/\n```\n\n**What good output looks like:**\n- pLDDT: > 85 (mean, on 0-100 scale) or > 0.85 (normalized)\n- pTM: > 0.70\n- ipTM: > 0.50 for complexes\n- PAE_interface: < 10\n\n## Decision tree\n\n```\nShould I use AlphaFold?\n│\n├─ What are you predicting?\n│  ├─ Single protein → ESMFold (faster)\n│  ├─ Protein-protein complex → AlphaFold/ColabFold ✓\n│  ├─ Protein + ligand → Chai or Boltz\n│  └─ Batch of sequences → ColabFold ✓\n│\n├─ What do you need?\n│  ├─ Highest accuracy → AlphaFold/ColabFold ✓\n│  ├─ Fast screening → ESMFold\n│  └─ MSA-free prediction → Chai or ESMFold\n│\n└─ Which AF2 option?\n   ├─ Local installation → Full control, slow setup\n   ├─ ColabFold → Easier, MSA server\n   └─ Modal → Recommended for batch\n```\n\n## Typical performance\n\n| Campaign Size | Time (A100) | Cost (Modal) | Notes |\n|---------------|-------------|--------------|-------|\n| 100 complexes | 1-2h | ~$8 | With MSA server |\n| 500 complexes | 5-10h | ~$40 | Standard campaign |\n| 1000 complexes | 10-20h | ~$80 | Large campaign |\n\n**Per-complex**: ~30-60s with MSA server.\n\n---\n\n## Verify\n\n```bash\nfind output -name \"ranked_0.pdb\" | wc -l  # Should match input count\n```\n\n---\n\n## Troubleshooting\n\n**Low pLDDT regions**: May indicate disorder or poor design\n**Low ipTM**: Interface not confident, check hotspots\n**High PAE off-diagonal**: Chains may not interact\n**OOM errors**: Use ColabFold with MSA server instead\n\n### Error interpretation\n\n| Error | Cause | Fix |\n|-------|-------|-----|\n| `RuntimeError: CUDA out of memory` | Sequence too long | Use A100 or split prediction |\n| `KeyError: 'iptm'` | Running monomer on complex | Use multimer preset |\n| `FileNotFoundError: database` | Missing MSA databases | Use ColabFold MSA server |\n| `TimeoutError` | MSA search slow | Reduce num_recycles |\n\n---\n\n**Next**: `protein-qc` for filtering and ranking.","tags":["alphafold","protein","design","skills","adaptyvbio","agent-skills","claude-code","protein-design","protein-engineering"],"capabilities":["skill","source-adaptyvbio","skill-alphafold","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/alphafold","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 (4,064 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":"2026-05-02T12:54:47.970Z","embedding":null,"createdAt":"2026-04-18T22:10:00.767Z","updatedAt":"2026-05-02T12:54:47.970Z","lastSeenAt":"2026-05-02T12:54:47.970Z","tsv":"'-01':134,135 '-10':385 '-100':288 '-2':376 '-20':165,393 '-60':402 '/../docs/installation.md':74 '/deepmind/alphafold.git':116 '0':287 '0.50':296 '0.62':271 '0.70':294 '0.78':269 '0.85':291 '1':13,82,164,375 '1/5':255 '10':301,392 '100':373 '1000':390 '100gb':62 '11.0':51 '12.0':52 '2':19,108 '2026':133 '3':26,137,163 '3.10':49 '3.8':48 '30':401 '32gb':55,59 '3600s':105 '4':33 '40':387 '40gb':56,103 '5':384 '5/5':259 '500':382 '500gb':63 '64gb':60 '8':378 '8.5':274 '80':395 '85':284 '87.3':266 'a100':57,102,369,467 'accuraci':335 'af2':348 'align':234 'alphafold':1,118,307 'alphafold/colabfold':320,336 'alphafold2':6,40 'bash':87,111,143,408 'batch':326,363 'best':188,192 'binder':22 'binder-target':21 'biomod':80,89 'boltz':325 'calcul':27 'campaign':366,389,397 'caus':456 'cd':88,117 'chai':323,344 'chain':142,441 'check':434 'clone':113 'colabfold':83,329,356,448,486 'complex':24,298,319,374,383,391,400,476 'complex.fasta':244 'confid':28,194,433 'consist':36 'control':353 'correct':18 'cost':370 'count':418 'cuda':50,459 'cutoff':176 'databas':65,481,484 'date':132,170 'dd':174 'decis':302 'default':106,152 'descript':154 'design':4,15,39,428 'diagon':440 'dir':99,125 'disk':61 'disord':425 'easier':357 'error':235,446,453,455 'esmfold':138,314,339,346 'extract':205 'f':215,218 'faa':95 'fast':139,337 'fasta':121,242 'faster':315 'featur':204 'features.pkl':202 'file':201 'filenotfounderror':480 'filter':501 'find':409 'first':69 'fix':457 'fold':17 'format':185 'free':342 'full':197,352 'git':112 'github.com':115 'github.com/deepmind/alphafold.git':114 'good':279 'gpu':53,101 'guid':73 'h':377,386,394 'high':436 'highest':334 'hotspot':435 'import':208 'indic':424 'info':248,252,256,260 'input':94,203,417 'input-faa':93 'instal':72,110,351 'instead':452 'interact':444 'interfac':273,300,431 'interpret':454 'iptm':32,225,227,270,295,430,472 'iter':167 'key':149 'keyerror':471 'l':414 'larg':396 'ligand':322 'like':282 'local':109,350 'long':465 'look':281 'low':420,429 'match':416 'max':130,168 'may':423,442 'mean':267,285 'memori':462 'metric':29,206 'minimum':45 'miss':482 'mktayiakqrqisfvk':148 'mm':173 'modal':78,90,144,360,371 'modal_colabfold.py':92 'modal_esmfold.py':146 'model':127,155,159,189,245,254,258 'monom':129,157,474 'monomer/multimer':158 'msa':200,250,341,358,380,405,450,483,487,490 'msa-fre':340 'msas':199 'multim':86,229,247,478 'name':411 'need':333 'next':496 'none':228 'normal':292 'note':372 'num':161,494 'off-diagon':438 'oom':445 'open':211 'option':81,107,136,153,349 'out-dir':97 'output':100,124,126,184,186,237,277,280,410 'pae':231,272,299,437 'paramet':150,151 'path':122,243 'per':399 'per-complex':398 'perform':365 'pickl':209 'pickle.load':217 'plddt':30,219,221,265,283,421 'poor':427 'predict':8,20,233,311,343,470 'prerequisit':43 'preset':128,156,246,479 'protein':3,313,317,318,321,498 'protein-protein':316 'protein-qc':497 'ptm':31,222,224,268,293 'python':47,119,207,240 'qc':499 'query.fasta':123 'ram':58 'rank':503 'ranked_0.pdb':187,264,412 'ranked_1.pdb':190 'ranking_debug.json':193 'rb':213 'recommend':46,84,361 'recycl':162,166,495 'reduc':493 'region':422 'relax':261 'requir':44 'result':198,216,220,223,232,263 'result.get':226 'result_model_1.pkl':196,212 'run':68,91,145,239,249,253,257,473 'run_alphafold.py':120,241 'runtimeerror':458 'sampl':236 'save':275 'scale':289 'score':195 'screen':338 'search':183,251,491 'second':191 'see':71 'self':35 'self-consist':34 'sequenc':16,147,328,463 'sequences.fasta':96 'server':359,381,406,451,488 'set':76 'setup':355 'singl':141,312 'single-chain':140 'size':367 'skill':11 'skill-alphafold' 'slow':354,492 'source-adaptyvbio' 'split':469 'standard':388 'structur':7,25,41,262 'success':238 'target':23 'templat':131,169,175,178,182 'time':70,368 'timeout':104 'timeouterror':489 'topic-agent-skills' 'topic-claude-code' 'topic-protein-design' 'topic-protein-engineering' 'tree':303 'troubleshoot':419 'true':179 'true/false':180 'type':160 'typic':364 'use':5,9,177,181,306,447,466,477,485 'valid':2,14,37,42 'verifi':407 'vram':54 'wc':413 'yyyi':172 'yyyy-mm-dd':171","prices":[{"id":"c9df0f04-0c3a-4e04-8acd-9b10bd9793b3","listingId":"4ae38127-c840-44be-92e1-3255b9691cdb","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":"2026-04-18T22:10:00.767Z"}],"sources":[{"listingId":"4ae38127-c840-44be-92e1-3255b9691cdb","source":"github","sourceId":"adaptyvbio/protein-design-skills/alphafold","sourceUrl":"https://github.com/adaptyvbio/protein-design-skills/tree/main/skills/alphafold","isPrimary":false,"firstSeenAt":"2026-04-18T22:10:00.767Z","lastSeenAt":"2026-05-02T12:54:47.970Z"}],"details":{"listingId":"4ae38127-c840-44be-92e1-3255b9691cdb","quickStartSnippet":null,"exampleRequest":null,"exampleResponse":null,"schema":null,"openapiUrl":null,"agentsTxtUrl":null,"citations":[],"useCases":[],"bestFor":[],"notFor":[],"kindDetails":{"org":"adaptyvbio","slug":"alphafold","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":"3a114fb433df98956d65762fb62c899241e23eed","skill_md_path":"skills/alphafold/SKILL.md","default_branch":"main","skill_tree_url":"https://github.com/adaptyvbio/protein-design-skills/tree/main/skills/alphafold"},"layout":"multi","source":"github","category":"protein-design-skills","frontmatter":{"name":"alphafold","license":"MIT","description":"Validate protein designs using AlphaFold2 structure prediction. Use this skill when: (1) Validating designed sequences fold correctly, (2) Predicting binder-target complex structures, (3) Calculating confidence metrics (pLDDT, pTM, ipTM), (4) Self-consistency validation of designs, (5) Multi-chain complex prediction with AlphaFold-Multimer.  For faster single-chain prediction, use esm. For QC thresholds, use protein-qc."},"skills_sh_url":"https://skills.sh/adaptyvbio/protein-design-skills/alphafold"},"updatedAt":"2026-05-02T12:54:47.970Z"}}