[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6DP31T6QBNE0CJJJ6C6VB8D":3,"skill-same:01M6DP31T6QBNE0CJJJ6C6VB8D":61},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":34,"linkOnly":35,"markdownContent":36,"markdownBody":37,"frontMatter":38,"refName":42,"commitSha":43,"skillDirectory":5,"forksCount":14,"repositoryHtmlUrl":44,"repositoryHomepageUrl":6,"ownerHtmlUrl":45,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":46,"install":47,"resources":60},"01M6DP31T6QBNE0CJJJ6C6VB8D","ipsae",null,"Binder design ranking using ipSAE (interprotein Score from Aligned Errors). Use this skill when: (1) Ranking binder designs for experimental testing, (2) Filtering BindCraft or RFdiffusion outputs, (3) Comparing AF2\u002FAF3\u002FBoltz predictions, (4) Predicting binding success rates, (5) Need better ranking than ipTM or iPAE.\nFor structure prediction, use chai or alphafold. For QC thresholds, use protein-qc.","huamu668\u002Fclawhub-openclaw-medical","huamu668","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F202473531?v=4","01M6DK1V7D7VP9PJ06Z1VXVK13","huamu","01M6DK1VBYG1BX5FT154AFZHAS",0,"MIT","https:\u002F\u002Fgithub.com\u002Fhuamu668\u002Fclawhub-openclaw-medical\u002Fblob\u002Fmain\u002Fipsae\u002FSKILL.md",[],[5,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33],"testing","design","Evaluation","filtering","use","scoring","structure","alphafold","binder","bindcraft","binder design","binding","boltz","chai","ranking","2026-11-10 10:12:30",false,"---\nname: ipsae\ndescription: >\n  Binder design ranking using ipSAE (interprotein Score from Aligned Errors).\n  Use this skill when: (1) Ranking binder designs for experimental testing,\n  (2) Filtering BindCraft or RFdiffusion outputs,\n  (3) Comparing AF2\u002FAF3\u002FBoltz predictions,\n  (4) Predicting binding success rates,\n  (5) Need better ranking than ipTM or iPAE.\n\n  For structure prediction, use chai or alphafold.\n  For QC thresholds, use protein-qc.\nlicense: MIT\ncategory: evaluation\ntags: [ranking, scoring, binding]\n---\n\n# ipSAE Binder Ranking\n\n## Prerequisites\n\n| Requirement | Minimum | Recommended |\n|-------------|---------|-------------|\n| Python | 3.8+ | 3.10 |\n| NumPy | 1.20+ | Latest |\n| RAM | 8GB | 16GB |\n\n## Overview\n\nipSAE (interprotein Score from Aligned Errors) is a scoring function for ranking protein-protein interactions predicted by AlphaFold2, AlphaFold3, and Boltz1. It outperforms ipTM and iPAE for binder design ranking with **1.4x higher precision** in identifying true binders.\n\n**Paper**: [What's wrong with AlphaFold's ipTM score](https:\u002F\u002Fwww.biorxiv.org\u002Fcontent\u002F10.1101\u002F2025.02.10.637595v2)\n\n## How to run\n\n### Installation\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002FDunbrackLab\u002FIPSAE.git\ncd IPSAE\npip install numpy\n```\n\n### AlphaFold2\n```bash\npython ipsae.py scores_rank_001.json unrelaxed_rank_001.pdb 15 15\n```\n\n### AlphaFold3\n```bash\npython ipsae.py fold_model_full_data_0.json fold_model_0.cif 10 10\n```\n\n### Boltz1\n```bash\npython ipsae.py pae_model_0.npz model_0.cif 10 10\n```\n\n## Key parameters\n\n| Parameter | Description | Recommended |\n|-----------|-------------|-------------|\n| PAE file | JSON (AF2\u002FAF3) or NPZ (Boltz) | Match predictor |\n| Structure file | PDB or CIF structure | Match PAE |\n| PAE cutoff | Threshold for contacts | 10-15 |\n| Distance cutoff | Max CA-CA distance (A) | 10-15 |\n\n## Output format\n\nTwo output files are generated:\n\n**Chain-pair scores** (`_chains.csv`):\n```\nchain_A,chain_B,ipSAE_min,pDockQ,pDockQ2,LIS,n_contacts,interface_dist\nA,B,0.72,0.65,0.58,0.45,42,8.5\n```\n\n**Residue-level scores** (`_residues.csv`):\n```\nchain,resnum,pSAE,pLDDT\nA,45,0.85,92.3\nA,67,0.78,88.1\n```\n\n## Sample output\n\n### Successful run\n```\n$ python ipsae.py scores_rank_001.json design_0.pdb 10 10\nProcessing design_0...\nFound 2 chains: A, B\nComputing ipSAE scores...\n\nResults written to:\n  design_0_chains.csv\n  design_0_residues.csv\n\nSummary:\n  ipSAE_min: 0.72\n  pDockQ: 0.65\n  LIS: 0.45\n  Interface contacts: 42\n```\n\n**What good output looks like:**\n- ipSAE_min > 0.61 (primary filter)\n- pDockQ > 0.5 (supporting metric)\n- Reasonable number of interface contacts (20-100)\n\n## Decision tree\n\n```\nShould I use ipSAE?\n│\n├─ What are you ranking?\n│  ├─ Designed binders → ipSAE ✓\n│  ├─ Natural complexes → ipTM is fine\n│  └─ Single proteins → Not applicable\n│\n├─ What predictor did you use?\n│  ├─ AlphaFold2 → ipSAE ✓\n│  ├─ AlphaFold3 → ipSAE ✓\n│  ├─ Boltz1 → ipSAE ✓\n│  ├─ Chai → ipSAE (use PAE output)\n│  └─ ESMFold → Not applicable (no PAE)\n│\n└─ Why ipSAE over ipTM?\n   ├─ Different length constructs → ipSAE ✓\n   ├─ Designs with disordered regions → ipSAE ✓\n   └─ Standard complexes → Either works\n```\n\n## Recommended thresholds\n\n| Metric | Standard | Stringent | Use Case |\n|--------|----------|-----------|----------|\n| ipSAE_min | > 0.61 | > 0.70 | Primary filter |\n| LIS | > 0.35 | > 0.45 | Interface quality |\n| pDockQ | > 0.5 | > 0.6 | Supporting |\n\n## Batch processing\n\n```python\nimport subprocess\nimport os\nfrom pathlib import Path\n\ndef score_designs(pae_dir, struct_dir, output_dir):\n    \"\"\"Score all designs in a directory.\"\"\"\n    Path(output_dir).mkdir(exist_ok=True)\n\n    for pae_file in Path(pae_dir).glob(\"*_scores*.json\"):\n        name = pae_file.stem.replace(\"_scores_rank_001\", \"\")\n        struct_file = Path(struct_dir) \u002F f\"{name}.pdb\"\n\n        if struct_file.exists():\n            subprocess.run([\n                \"python\", \"ipsae.py\",\n                str(pae_file),\n                str(struct_file),\n                \"10\", \"10\"\n            ])\n```\n\n---\n\n## Verify\n\n```bash\nls *_chains.csv | wc -l  # Should match number of predictions\n```\n\n---\n\n## Troubleshooting\n\n**Low scores for good designs**: Check PAE\u002Fdistance cutoffs\n**Missing output**: Verify PAE file format matches predictor\n**Inconsistent scores**: Use same cutoffs across all designs\n\n### Error interpretation\n\n| Error | Cause | Fix |\n|-------|-------|-----|\n| `KeyError: 'pae'` | Wrong PAE format | Check if AF2\u002FAF3\u002FBoltz format |\n| `FileNotFoundError` | Structure not found | Verify file paths |\n| `ValueError: no contacts` | No interface detected | Check chain IDs, reduce cutoffs |\n\n---\n\n**Next**: Select top designs (ipSAE_min > 0.61) → experimental validation.\n","# ipSAE Binder Ranking\n\n## Prerequisites\n\n| Requirement | Minimum | Recommended |\n|-------------|---------|-------------|\n| Python | 3.8+ | 3.10 |\n| NumPy | 1.20+ | Latest |\n| RAM | 8GB | 16GB |\n\n## Overview\n\nipSAE (interprotein Score from Aligned Errors) is a scoring function for ranking protein-protein interactions predicted by AlphaFold2, AlphaFold3, and Boltz1. It outperforms ipTM and iPAE for binder design ranking with **1.4x higher precision** in identifying true binders.\n\n**Paper**: [What's wrong with AlphaFold's ipTM score](https:\u002F\u002Fwww.biorxiv.org\u002Fcontent\u002F10.1101\u002F2025.02.10.637595v2)\n\n## How to run\n\n### Installation\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002FDunbrackLab\u002FIPSAE.git\ncd IPSAE\npip install numpy\n```\n\n### AlphaFold2\n```bash\npython ipsae.py scores_rank_001.json unrelaxed_rank_001.pdb 15 15\n```\n\n### AlphaFold3\n```bash\npython ipsae.py fold_model_full_data_0.json fold_model_0.cif 10 10\n```\n\n### Boltz1\n```bash\npython ipsae.py pae_model_0.npz model_0.cif 10 10\n```\n\n## Key parameters\n\n| Parameter | Description | Recommended |\n|-----------|-------------|-------------|\n| PAE file | JSON (AF2\u002FAF3) or NPZ (Boltz) | Match predictor |\n| Structure file | PDB or CIF structure | Match PAE |\n| PAE cutoff | Threshold for contacts | 10-15 |\n| Distance cutoff | Max CA-CA distance (A) | 10-15 |\n\n## Output format\n\nTwo output files are generated:\n\n**Chain-pair scores** (`_chains.csv`):\n```\nchain_A,chain_B,ipSAE_min,pDockQ,pDockQ2,LIS,n_contacts,interface_dist\nA,B,0.72,0.65,0.58,0.45,42,8.5\n```\n\n**Residue-level scores** (`_residues.csv`):\n```\nchain,resnum,pSAE,pLDDT\nA,45,0.85,92.3\nA,67,0.78,88.1\n```\n\n## Sample output\n\n### Successful run\n```\n$ python ipsae.py scores_rank_001.json design_0.pdb 10 10\nProcessing design_0...\nFound 2 chains: A, B\nComputing ipSAE scores...\n\nResults written to:\n  design_0_chains.csv\n  design_0_residues.csv\n\nSummary:\n  ipSAE_min: 0.72\n  pDockQ: 0.65\n  LIS: 0.45\n  Interface contacts: 42\n```\n\n**What good output looks like:**\n- ipSAE_min > 0.61 (primary filter)\n- pDockQ > 0.5 (supporting metric)\n- Reasonable number of interface contacts (20-100)\n\n## Decision tree\n\n```\nShould I use ipSAE?\n│\n├─ What are you ranking?\n│  ├─ Designed binders → ipSAE ✓\n│  ├─ Natural complexes → ipTM is fine\n│  └─ Single proteins → Not applicable\n│\n├─ What predictor did you use?\n│  ├─ AlphaFold2 → ipSAE ✓\n│  ├─ AlphaFold3 → ipSAE ✓\n│  ├─ Boltz1 → ipSAE ✓\n│  ├─ Chai → ipSAE (use PAE output)\n│  └─ ESMFold → Not applicable (no PAE)\n│\n└─ Why ipSAE over ipTM?\n   ├─ Different length constructs → ipSAE ✓\n   ├─ Designs with disordered regions → ipSAE ✓\n   └─ Standard complexes → Either works\n```\n\n## Recommended thresholds\n\n| Metric | Standard | Stringent | Use Case |\n|--------|----------|-----------|----------|\n| ipSAE_min | > 0.61 | > 0.70 | Primary filter |\n| LIS | > 0.35 | > 0.45 | Interface quality |\n| pDockQ | > 0.5 | > 0.6 | Supporting |\n\n## Batch processing\n\n```python\nimport subprocess\nimport os\nfrom pathlib import Path\n\ndef score_designs(pae_dir, struct_dir, output_dir):\n    \"\"\"Score all designs in a directory.\"\"\"\n    Path(output_dir).mkdir(exist_ok=True)\n\n    for pae_file in Path(pae_dir).glob(\"*_scores*.json\"):\n        name = pae_file.stem.replace(\"_scores_rank_001\", \"\")\n        struct_file = Path(struct_dir) \u002F f\"{name}.pdb\"\n\n        if struct_file.exists():\n            subprocess.run([\n                \"python\", \"ipsae.py\",\n                str(pae_file),\n                str(struct_file),\n                \"10\", \"10\"\n            ])\n```\n\n---\n\n## Verify\n\n```bash\nls *_chains.csv | wc -l  # Should match number of predictions\n```\n\n---\n\n## Troubleshooting\n\n**Low scores for good designs**: Check PAE\u002Fdistance cutoffs\n**Missing output**: Verify PAE file format matches predictor\n**Inconsistent scores**: Use same cutoffs across all designs\n\n### Error interpretation\n\n| Error | Cause | Fix |\n|-------|-------|-----|\n| `KeyError: 'pae'` | Wrong PAE format | Check if AF2\u002FAF3\u002FBoltz format |\n| `FileNotFoundError` | Structure not found | Verify file paths |\n| `ValueError: no contacts` | No interface detected | Check chain IDs, reduce cutoffs |\n\n---\n\n**Next**: Select top designs (ipSAE_min > 0.61) → experimental validation.\n",{"name":5,"tags":39,"license":15,"category":40,"description":41},[33,24,30],"evaluation","Binder design ranking using ipSAE (interprotein Score from Aligned Errors). Use this skill when: (1) Ranking binder designs for experimental testing, (2) Filtering BindCraft or RFdiffusion outputs, (3) Comparing AF2\u002FAF3\u002FBoltz predictions, (4) Predicting binding success rates, (5) Need better ranking than ipTM or iPAE.\nFor structure prediction, use chai or alphafold. For QC thresholds, use protein-qc.\n","main","d3998984163a67a891ee8397e35adf79b14e872d","https:\u002F\u002Fgithub.com\u002Fhuamu668\u002Fclawhub-openclaw-medical","https:\u002F\u002Fgithub.com\u002Fhuamu668","2026-03-12 11:49:42",{"source":8,"skillName":5,"commands":48,"prompt":58,"zipAvailable":59},[49,52,55],{"tool":50,"command":51},"npx","npx skills add huamu668\u002Fclawhub-openclaw-medical --skill ipsae",{"tool":53,"command":54},"bunx","bunx skills add huamu668\u002Fclawhub-openclaw-medical --skill ipsae",{"tool":56,"command":57},"pnpm","pnpm dlx skills add huamu668\u002Fclawhub-openclaw-medical --skill ipsae","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fhuamu668\u002Fclawhub-openclaw-medical\u002Fblob\u002Fmain\u002Fipsae\u002FSKILL.md",true,[],[62,82,112,132,146,174],{"skillNo":63,"name":64,"nameZh":6,"description":65,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":66,"categories":67,"keywords":68,"publishedAt":34},"01M6DR3JGAEX5G5FR7TK3D8NS8","zarr-python","Chunked N-D arrays for cloud storage. Compressed arrays, parallel I\u002FO, S3\u002FGCS integration, NumPy\u002FDask\u002FXarray compatible, for large-scale scientific computing pipelines.","https:\u002F\u002Fgithub.com\u002Fhuamu668\u002Fclawhub-openclaw-medical\u002Fblob\u002Fmain\u002Fzarr-python\u002FSKILL.md",[],[69,70,71,72,73,74,75,76,77,78,79,80,81],"python","zarr python","storage","scientific computing","numpy","arrays","pipelines","parallel","dask","cloud","gcs","s3","integration",{"skillNo":83,"name":84,"nameZh":6,"description":85,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":86,"categories":87,"keywords":93,"publishedAt":34},"01M6DR101MEDE5XT2ZZ5VBEAA5","vaex","Use this skill for processing and analyzing large tabular datasets (billions of rows) that exceed available RAM. Vaex excels at out-of-core DataFrame operations, lazy evaluation, fast aggregations, efficient visualization of big data, and machine learning on large datasets. Apply when users need to work with large CSV\u002FHDF5\u002FArrow\u002FParquet files, perform fast statistics on massive datasets, create visualizations of big data, or build ML pipelines that do not fit in memory.","https:\u002F\u002Fgithub.com\u002Fhuamu668\u002Fclawhub-openclaw-medical\u002Fblob\u002Fmain\u002Fvaex\u002FSKILL.md",[88],{"categoryNo":89,"categoryName":90,"taxonomyType":91,"taxonomyTypeDesc":92},"01JBSEEDCAT000000000000013","分析与可视化","SCENARIO","使用场景",[84,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,23,110,111],"machine learning","in memory","learning","csv","visualization","create","core","ml","big data","memory","visualizations","fit","statistics","do","build","files","operations","work",{"skillNo":113,"name":114,"nameZh":6,"description":115,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":116,"categories":117,"keywords":118,"publishedAt":34},"01M6DQVMB0GBKF2W44FKXVD1P7","torch-geometric","Graph Neural Networks (PyG). Node\u002Fgraph classification, link prediction, GCN, GAT, GraphSAGE, heterogeneous graphs, molecular property prediction, for geometric deep learning.","https:\u002F\u002Fgithub.com\u002Fhuamu668\u002Fclawhub-openclaw-medical\u002Fblob\u002Fmain\u002Ftorch-geometric\u002FSKILL.md",[],[119,96,120,121,122,123,124,125,126,127,128,129,130,131],"torch geometric","node","Neural","link","gat","molecular","graphs","deep","networks","graph","neural networks","deep learning","classification",{"skillNo":133,"name":134,"nameZh":6,"description":135,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":136,"categories":137,"keywords":138,"publishedAt":34},"01M6DQH2GZPZKEQ140GRCJ9746","tiledbvcf","Efficient storage and retrieval of genomic variant data using TileDB. Scalable VCF\u002FBCF ingestion, incremental sample addition, compressed storage, parallel queries, and export capabilities for population genomics.","https:\u002F\u002Fgithub.com\u002Fhuamu668\u002Fclawhub-openclaw-medical\u002Fblob\u002Fmain\u002Ftiledbvcf\u002FSKILL.md",[],[134,139,140,141,142,143,144,76,145,71],"data","genomics","sample","queries","ingestion","export","capabilities",{"skillNo":147,"name":148,"nameZh":6,"description":149,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":150,"categories":151,"keywords":160,"publishedAt":34},"01M6DQD9G958EM6NHJ994D0SK1","solublempnn","Solubility-optimized protein sequence design using SolubleMPNN. Use this skill when: (1) Designing for E. coli expression, (2) Optimizing solubility of designed proteins, (3) Reducing aggregation propensity, (4) Need high-yield expression, (5) Avoiding inclusion body formation.\nFor standard design, use proteinmpnn. For ligand-aware design, use ligandmpnn.","https:\u002F\u002Fgithub.com\u002Fhuamu668\u002Fclawhub-openclaw-medical\u002Fblob\u002Fmain\u002Fsolublempnn\u002FSKILL.md",[152,155],{"categoryNo":153,"categoryName":154,"taxonomyType":91,"taxonomyTypeDesc":92},"01JBSEEDCAT000000000000015","视觉设计",{"categoryNo":156,"categoryName":157,"taxonomyType":158,"taxonomyTypeDesc":159},"01JBSEEDCAT000000000000032","设计师","PROFESSION","适用职业",[148,161,162,23,163,164,165,20,166,167,168,169,170,171,172,173],"inclusion","yield","modal","Tools","aggregation","proteins","solubility","proteinmpnn","ligandmpnn","ligand-aware","inverse-folding","expression","sequence-design",{"skillNo":175,"name":176,"nameZh":6,"description":177,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":178,"licenseFileUrl":6,"sourceHtmlUrl":179,"categories":180,"keywords":181,"publishedAt":34},"01M6DQ8W91A93MGGCEEW949HBC","scvelo","RNA velocity analysis with scVelo. Estimate cell state transitions from unspliced\u002Fspliced mRNA dynamics, infer trajectory directions, compute latent time, and identify driver genes in single-cell RNA-seq data. Complements Scanpy\u002FscVI-tools for trajectory inference.","BSD-3-Clause","https:\u002F\u002Fgithub.com\u002Fhuamu668\u002Fclawhub-openclaw-medical\u002Fblob\u002Fmain\u002Fscvelo\u002FSKILL.md",[],[176,182,183,184,185,186,187,188,189,190,191,164,139,192,193],"RNA","single-cell","infer","cell","transitions","time","compute","velocity","scanpy","analysis","state","inference"]