[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7AWVNGRA9P7G8R98PVG50R8":3,"skill-same:01M7AWVNGRA9P7G8R98PVG50R8":55},{"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":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":33,"commitSha":34,"skillDirectory":35,"forksCount":14,"repositoryHtmlUrl":36,"repositoryHomepageUrl":6,"ownerHtmlUrl":37,"ownerBlogUrl":38,"ownerDescription":39,"updatedAt":40,"install":41,"resources":54},"01M7AWVNGRA9P7G8R98PVG50R8","proteinmpnn",null,"Inverse-fold a protein backbone (PDB structure) into amino-acid sequence with ProteinMPNN (Dauparas et al. 2022, github.com\u002Fdauparas\u002FProteinMPNN). Reach for this skill to run sequence design on RFdiffusion backbones, to redesign one chain of a PDB while holding interface residues fixed, or to generate a temperature-swept set of sequences for downstream folding.","flyme2bluemoon\u002Fskills","flyme2bluemoon","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F37197710?v=4","01M7AWSKCXRPGZSNS26MTB5YR4","Matthew Shen","01M7AWSKE11G83QK1885PSGJ3K",0,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fflyme2bluemoon\u002Fskills\u002Fblob\u002Fmain\u002Fskills-examples\u002Fclaude-science-skills\u002Fproteinmpnn\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: proteinmpnn\ndescription: >\n  Inverse-fold a protein backbone (PDB structure) into amino-acid sequence with\n  ProteinMPNN (Dauparas et al. 2022, github.com\u002Fdauparas\u002FProteinMPNN). Reach\n  for this skill to run sequence design on RFdiffusion backbones, to redesign\n  one chain of a PDB while holding interface residues fixed, or to generate a\n  temperature-swept set of sequences for downstream folding.\nlicense: Apache-2.0\nmetadata:\n  category: biomodels\n  display-name: ProteinMPNN\n  # github.com\u002Fdauparas\u002FProteinMPNN\u002Fblob\u002Fmain\u002FLICENSE: MIT (© 2022 Justas\n  # Dauparas). verified 2026-06-30\n  third_party:\n    - kind: weights\n      name: ProteinMPNN\n      license: MIT\n      terms_url: https:\u002F\u002Fgithub.com\u002Fdauparas\u002FProteinMPNN\u002Fblob\u002Fmain\u002FLICENSE\n---\n\n# ProteinMPNN\n\nProteinMPNN is the default inverse-folding step in the binder pipeline: a\nmessage-passing network that sees backbone geometry only, so it is the right\nchoice when the design surface is protein–protein and the wrong one as soon as\na ligand, nucleic acid, or metal is part of the interface — `ligandmpnn` adds\nthose atoms to the graph with a near-identical CLI, and `solublempnn` swaps in\nweights trained on soluble structures for an expression-biased prior. Code and\nweights are MIT (github.com\u002Fdauparas\u002FProteinMPNN). The model is small enough\nto run on CPU — for a handful of sequences on one backbone that is seconds and\nusually faster than dispatching a remote job; a GPU helps for batched\ncampaigns (hundreds of backbones or large `--num_seq_per_target`). Either way\nthe repo is cloned in-job — there is no PyPI dist and the checkpoints are\nbundled in the repo.\n\n## Running it\n\n```bash\npip install torch numpy   # if not already present\ngit clone --depth 1 https:\u002F\u002Fgithub.com\u002Fdauparas\u002FProteinMPNN.git proteinmpnn\ncd proteinmpnn\npython protein_mpnn_run.py \\\n  --pdb_path backbone.pdb --pdb_path_chains \"A\" \\\n  --out_folder out --num_seq_per_target 16 --sampling_temp \"0.1\"\n```\n\nTwo flags trip almost everyone the first time. `--sampling_temp` is parsed as a\nspace-separated string so one run can sweep several temperatures; a single\nvalue needs no quoting, but a multi-value sweep must be quoted\n(`\"0.1 0.2 0.3\"`), and commas never split — `\"0.1,0.2\"` fails the float cast. `--pdb_path_chains` is also space-separated inside\none quoted argument (`\"A B\"`); a comma is kept as part of the chain ID.\n\nDesigns land in `out\u002Fseqs\u002F\u003Cpdb_stem>.fa`. The first record is the input\nsequence; each design header carries `score=` (mean negative log-likelihood —\nlower is more confident), `global_score=`, and `seq_recovery=`. ProteinMPNN\nwrites sequences only — it does not thread them back onto the backbone; if you\nneed designed-sequence PDBs, the `ligandmpnn` runner writes them to\n`backbones\u002F` automatically and accepts `--model_type protein_mpnn` for the\nsame weights.\n\n## A flat chain map in `--fixed_positions_jsonl` silently redesigns every residue\n\n`--fixed_positions_jsonl` expects one JSON object per line keyed by the **PDB\nstem** first, then chain, then a list of 1-indexed residue numbers:\n`{\"backbone\": {\"A\": [10, 11, 12], \"B\": []}}`. Passing the inner\n`{\"A\": [...]}` directly — the obvious guess — is silently treated as \"no PDB\nmatched,\" and every position is redesigned. The bundled\n`helper_scripts\u002Fmake_fixed_positions_dict.py` writes the correct shape from a\nchain and range string and is worth the extra call; the same outer-stem rule\napplies to `--chain_id_jsonl` and `--tied_positions_jsonl`.\n\n## Checkpoints — which one to pick\n\n| `--model_name` | training noise | use |\n|---|---|---|\n| `v_48_002` | 0.02 Å | highest recovery; close-to-native redesigns |\n| `v_48_020` (default) | 0.20 Å | de novo backbones — tolerates RFdiffusion imperfection |\n| `v_48_030` | 0.30 Å | very rough backbones; lowest recovery |\n| `--use_soluble_model` | — | swaps to the soluble-trained set; see `solublempnn` |\n\n## Errors worth recognizing\n\n| You see | It means \u002F do this |\n|---|---|\n| `KeyError: 'A'` | Chain letter not in the PDB — `grep '^ATOM' file.pdb \\| cut -c22 \\| sort -u` to see what is. |\n| `JSONDecodeError` on a `*_jsonl` flag | The flag wants a file path, not inline JSON; write the file first. |\n| All positions redesigned despite `--fixed_positions_jsonl` | Outer PDB-stem key missing — see the gotcha above. |\n| `ModuleNotFoundError` for relative imports | Script run from the wrong cwd — `cd` into the cloned repo first; the imports are repo-relative. |\n\n---\n\n**Next:** fold the designs in complex with the target via `boltz`, `chai1`, or\n`esmfold2` and filter on ipTM.\n","# ProteinMPNN\n\nProteinMPNN is the default inverse-folding step in the binder pipeline: a\nmessage-passing network that sees backbone geometry only, so it is the right\nchoice when the design surface is protein–protein and the wrong one as soon as\na ligand, nucleic acid, or metal is part of the interface — `ligandmpnn` adds\nthose atoms to the graph with a near-identical CLI, and `solublempnn` swaps in\nweights trained on soluble structures for an expression-biased prior. Code and\nweights are MIT (github.com\u002Fdauparas\u002FProteinMPNN). The model is small enough\nto run on CPU — for a handful of sequences on one backbone that is seconds and\nusually faster than dispatching a remote job; a GPU helps for batched\ncampaigns (hundreds of backbones or large `--num_seq_per_target`). Either way\nthe repo is cloned in-job — there is no PyPI dist and the checkpoints are\nbundled in the repo.\n\n## Running it\n\n```bash\npip install torch numpy   # if not already present\ngit clone --depth 1 https:\u002F\u002Fgithub.com\u002Fdauparas\u002FProteinMPNN.git proteinmpnn\ncd proteinmpnn\npython protein_mpnn_run.py \\\n  --pdb_path backbone.pdb --pdb_path_chains \"A\" \\\n  --out_folder out --num_seq_per_target 16 --sampling_temp \"0.1\"\n```\n\nTwo flags trip almost everyone the first time. `--sampling_temp` is parsed as a\nspace-separated string so one run can sweep several temperatures; a single\nvalue needs no quoting, but a multi-value sweep must be quoted\n(`\"0.1 0.2 0.3\"`), and commas never split — `\"0.1,0.2\"` fails the float cast. `--pdb_path_chains` is also space-separated inside\none quoted argument (`\"A B\"`); a comma is kept as part of the chain ID.\n\nDesigns land in `out\u002Fseqs\u002F\u003Cpdb_stem>.fa`. The first record is the input\nsequence; each design header carries `score=` (mean negative log-likelihood —\nlower is more confident), `global_score=`, and `seq_recovery=`. ProteinMPNN\nwrites sequences only — it does not thread them back onto the backbone; if you\nneed designed-sequence PDBs, the `ligandmpnn` runner writes them to\n`backbones\u002F` automatically and accepts `--model_type protein_mpnn` for the\nsame weights.\n\n## A flat chain map in `--fixed_positions_jsonl` silently redesigns every residue\n\n`--fixed_positions_jsonl` expects one JSON object per line keyed by the **PDB\nstem** first, then chain, then a list of 1-indexed residue numbers:\n`{\"backbone\": {\"A\": [10, 11, 12], \"B\": []}}`. Passing the inner\n`{\"A\": [...]}` directly — the obvious guess — is silently treated as \"no PDB\nmatched,\" and every position is redesigned. The bundled\n`helper_scripts\u002Fmake_fixed_positions_dict.py` writes the correct shape from a\nchain and range string and is worth the extra call; the same outer-stem rule\napplies to `--chain_id_jsonl` and `--tied_positions_jsonl`.\n\n## Checkpoints — which one to pick\n\n| `--model_name` | training noise | use |\n|---|---|---|\n| `v_48_002` | 0.02 Å | highest recovery; close-to-native redesigns |\n| `v_48_020` (default) | 0.20 Å | de novo backbones — tolerates RFdiffusion imperfection |\n| `v_48_030` | 0.30 Å | very rough backbones; lowest recovery |\n| `--use_soluble_model` | — | swaps to the soluble-trained set; see `solublempnn` |\n\n## Errors worth recognizing\n\n| You see | It means \u002F do this |\n|---|---|\n| `KeyError: 'A'` | Chain letter not in the PDB — `grep '^ATOM' file.pdb \\| cut -c22 \\| sort -u` to see what is. |\n| `JSONDecodeError` on a `*_jsonl` flag | The flag wants a file path, not inline JSON; write the file first. |\n| All positions redesigned despite `--fixed_positions_jsonl` | Outer PDB-stem key missing — see the gotcha above. |\n| `ModuleNotFoundError` for relative imports | Script run from the wrong cwd — `cd` into the cloned repo first; the imports are repo-relative. |\n\n---\n\n**Next:** fold the designs in complex with the target via `boltz`, `chai1`, or\n`esmfold2` and filter on ipTM.\n",{"name":5,"license":15,"metadata":24,"description":32},{"category":25,"third_party":26,"display-name":29},"biomodels",[27],{"kind":28,"name":29,"license":30,"terms_url":31},"weights","ProteinMPNN","MIT","https:\u002F\u002Fgithub.com\u002Fdauparas\u002FProteinMPNN\u002Fblob\u002Fmain\u002FLICENSE","Inverse-fold a protein backbone (PDB structure) into amino-acid sequence with ProteinMPNN (Dauparas et al. 2022, github.com\u002Fdauparas\u002FProteinMPNN). Reach for this skill to run sequence design on RFdiffusion backbones, to redesign one chain of a PDB while holding interface residues fixed, or to generate a temperature-swept set of sequences for downstream folding.\n","main","9a28ca56870fad42cc4efa04a1d487c07a561ab1","skills-examples\u002Fclaude-science-skills\u002Fproteinmpnn","https:\u002F\u002Fgithub.com\u002Fflyme2bluemoon\u002Fskills","https:\u002F\u002Fgithub.com\u002Fflyme2bluemoon","https:\u002F\u002Fflyme2bluemoon.github.io","Computer Science and Biology @ McGill University | making computer do cool thing","2026-10-31 13:32:13",{"source":8,"skillName":5,"commands":42,"prompt":52,"zipAvailable":53},[43,46,49],{"tool":44,"command":45},"npx","npx skills add flyme2bluemoon\u002Fskills --skill proteinmpnn",{"tool":47,"command":48},"bunx","bunx skills add flyme2bluemoon\u002Fskills --skill proteinmpnn",{"tool":50,"command":51},"pnpm","pnpm dlx skills add flyme2bluemoon\u002Fskills --skill proteinmpnn","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fflyme2bluemoon\u002Fskills\u002Fblob\u002Fmain\u002Fskills-examples\u002Fclaude-science-skills\u002Fproteinmpnn\u002FSKILL.md",true,[],[56,63,70,77,84,91],{"skillNo":57,"name":58,"nameZh":6,"description":59,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":30,"licenseFileUrl":6,"sourceHtmlUrl":60,"categories":61,"keywords":62,"publishedAt":19},"01M7AWZ83KYJ9YSB4E5VQJ1K9B","unslop","Edit substantive human-facing prose to remove AI writing patterns while preserving meaning and tone.","https:\u002F\u002Fgithub.com\u002Fflyme2bluemoon\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fpstack\u002Funslop\u002FSKILL.md",[],[],{"skillNo":64,"name":65,"nameZh":6,"description":66,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":30,"licenseFileUrl":6,"sourceHtmlUrl":67,"categories":68,"keywords":69,"publishedAt":19},"01M7AWZ4FJPVNRMFFJC6JSHHKQ","pstack-principle-type-system-discipline","Apply when designing types, reviewing a function signature, or writing code in any statically-typed language. Make illegal states unrepresentable, brand semantic primitives, parse external data at boundaries, refuse to lie to the compiler, exhaust variants, derive from authoritative schemas.","https:\u002F\u002Fgithub.com\u002Fflyme2bluemoon\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fpstack\u002Fpstack-principle-type-system-discipline\u002FSKILL.md",[],[],{"skillNo":71,"name":72,"nameZh":6,"description":73,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":30,"licenseFileUrl":6,"sourceHtmlUrl":74,"categories":75,"keywords":76,"publishedAt":19},"01M7AWZ11K4KET3ASESNABPE52","pstack-principle-test-behavior-not-implementation","Apply when you write, change, or keep a test. Call the code the way its users do and assert the result they observe against a literal expected value. If the test would still pass when every imported function returns undefined, rewrite the assertion or delete the test.","https:\u002F\u002Fgithub.com\u002Fflyme2bluemoon\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fpstack\u002Fpstack-principle-test-behavior-not-implementation\u002FSKILL.md",[],[],{"skillNo":78,"name":79,"nameZh":6,"description":80,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":30,"licenseFileUrl":6,"sourceHtmlUrl":81,"categories":82,"keywords":83,"publishedAt":19},"01M7AWYWT47M5JG7XV6ZK6QSSX","pstack-principle-subtract-before-you-add","Apply when sequencing an addition, refactor, or rewrite. Remove dead weight, redundant validators, and stub references first, then build on the simpler base.","https:\u002F\u002Fgithub.com\u002Fflyme2bluemoon\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fpstack\u002Fpstack-principle-subtract-before-you-add\u002FSKILL.md",[],[],{"skillNo":85,"name":86,"nameZh":6,"description":87,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":30,"licenseFileUrl":6,"sourceHtmlUrl":88,"categories":89,"keywords":90,"publishedAt":19},"01M7AWYTNNHNGHB583NEV888BB","pstack-principle-sequence-verifiable-units","Sequence migrations or dependent changes around meaningful verification boundaries.","https:\u002F\u002Fgithub.com\u002Fflyme2bluemoon\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fpstack\u002Fpstack-principle-sequence-verifiable-units\u002FSKILL.md",[],[],{"skillNo":92,"name":93,"nameZh":6,"description":94,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":30,"licenseFileUrl":6,"sourceHtmlUrl":95,"categories":96,"keywords":97,"publishedAt":19},"01M7AWYQ2C15PV2W2V6ZVMK3ZM","pstack-principle-separate-before-serializing-shared-state","Apply when concurrent actors might write to the same file, branch, key, or state object. Eliminate the sharing first; serialize structurally only when one shared writer is a real invariant.","https:\u002F\u002Fgithub.com\u002Fflyme2bluemoon\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fpstack\u002Fpstack-principle-separate-before-serializing-shared-state\u002FSKILL.md",[],[]]