[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M74MFPSG3Y2ZB74ZXRHT2P51":3,"skill-same:01M74MFPSG3Y2ZB74ZXRHT2P51":57},{"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":28,"commitSha":29,"skillDirectory":30,"forksCount":31,"repositoryHtmlUrl":32,"repositoryHomepageUrl":33,"ownerHtmlUrl":34,"ownerBlogUrl":6,"ownerDescription":35,"updatedAt":36,"install":37,"resources":50},"01M74MFPSG3Y2ZB74ZXRHT2P51","scvi-tools",null,"Probabilistic single-cell RNA-seq with scvi-tools — scVI for a batch-corrected latent space, scANVI for semi-supervised label transfer, and Bayesian differential expression. Reach for this skill to integrate scRNA-seq batches, embed cells for clustering, transfer annotations from a reference onto a query, or score differentially expressed genes per cluster. For spatial deconvolution \u002F mapping use the cell2location, DestVI, or Tangram methods instead.","pengqianhan\u002FAI-Human-Research-OS","pengqianhan","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F4286531?v=4","01M74M3DT9HNZ6HPAK8TX69EX9","PengqianHan","01M74M3DTVSN2GXW0JH2W396FK",3,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fpengqianhan\u002FAI-Human-Research-OS\u002Fblob\u002Fmain\u002Fresearch-skills-hub\u002Fclaude-science-skills\u002Fskills\u002Fscvi-tools\u002FSKILL.md",[],[],"2026-11-11 15:11:14",false,"---\nname: scvi-tools\ndescription: >\n  Probabilistic single-cell RNA-seq with scvi-tools — scVI for a\n  batch-corrected latent space, scANVI for semi-supervised label transfer,\n  and Bayesian differential expression. Reach for this skill to integrate\n  scRNA-seq batches, embed cells for clustering, transfer annotations from a\n  reference onto a query, or score differentially expressed genes per cluster.\n  For spatial deconvolution \u002F mapping use the cell2location, DestVI, or\n  Tangram methods instead.\nlicense: Apache-2.0\nrequirements: [gpu]\nmetadata:\n  display-name: scvi-tools\n---\n\n# scvi-tools — scVI \u002F scANVI\n\nscvi-tools (Gayoso et al. 2022, github.com\u002Fscverse\u002Fscvi-tools, BSD-3-Clause)\nwraps a family\nof deep generative models for single-cell omics. The scRNA-seq core is **scVI**\n(unsupervised batch-corrected latent embedding) and **scANVI** (scVI + a\nclassifier head for semi-supervised cell-type label transfer). Both expect\n**raw integer UMI counts** and emit a low-dimensional `X_scVI` \u002F `X_scANVI`\nthat drops into the scanpy neighbors → leiden → umap pipeline.\n\n## How to run\n\n### scVI — batch-corrected latent space\n\n```python\nimport scanpy as sc\nimport scvi\n\nadata = sc.read_h5ad(\"dataset.h5ad\")\nadata.layers[\"counts\"] = adata.X.copy()          # preserve raw BEFORE any normalize\u002Flog1p\nsc.pp.normalize_total(adata); sc.pp.log1p(adata) # optional, for HVG \u002F plotting only\nsc.pp.highly_variable_genes(adata, n_top_genes=2000, batch_key=\"batch\", subset=True)\n\nscvi.model.SCVI.setup_anndata(adata, layer=\"counts\", batch_key=\"batch\")\nmodel = scvi.model.SCVI(adata, n_latent=30)\nmodel.train(max_epochs=200, early_stopping=True, accelerator=\"gpu\", devices=1)\n\nadata.obsm[\"X_scVI\"] = model.get_latent_representation()\nadata.layers[\"scvi_normalized\"] = model.get_normalized_expression(library_size=1e4)\n```\n\n### scANVI — label transfer from a partially-annotated reference\n\n```python\nlvae = scvi.model.SCANVI.from_scvi_model(\n    model, labels_key=\"cell_type\", unlabeled_category=\"Unknown\",\n)\nlvae.train(max_epochs=20, n_samples_per_label=100, accelerator=\"gpu\", devices=1)\n\nadata.obsm[\"X_scANVI\"] = lvae.get_latent_representation()\nadata.obs[\"pred_cell_type\"] = lvae.predict()\n```\n\n`accelerator=\"gpu\", devices=1` is the PyTorch-Lightning spelling; the legacy\n`use_gpu=` kwarg was **removed** in scvi-tools 1.x and now raises `TypeError`.\n\n## Differential expression\n\n```python\nde = model.differential_expression(\n    groupby=\"leiden\", group1=\"3\",   # group2=None → vs. all other cells\n    mode=\"change\", delta=0.25,\n)\ntop = de.sort_values(\"proba_de\", ascending=False).head(50)\n```\n\nFor one-vs-rest leave `group2` out — `\"rest\"` is scanpy's\n`rank_genes_groups` convention, not scvi-tools'; here `group2` is a literal\ncategory name and `\"rest\"` would match zero cells.\n\nscvi-tools ≥1.4 defaults to `mode=\"vanilla\"`, whose result columns are\nexactly:\n\n```\n['proba_m1', 'proba_m2', 'bayes_factor', 'scale1', 'scale2', 'raw_mean1',\n 'raw_mean2', 'non_zeros_proportion1', 'non_zeros_proportion2',\n 'raw_normalized_mean1', 'raw_normalized_mean2', 'comparison', 'group1',\n 'group2']\n```\n\n— no `lfc_*`, no `proba_de`, no `is_de_fdr_*`. **Pass `mode=\"change\"`** to\nget `lfc_mean` \u002F `lfc_median` \u002F `proba_de` \u002F `is_de_fdr_0.05`. Sort on\n`proba_de` (or on `bayes_factor` if you deliberately stayed in vanilla\nmode).\n\n## Output format\n\n| Key                            | What                                                   |\n| ------------------------------ | ------------------------------------------------------ |\n| `adata.obsm[\"X_scVI\"]`         | `n_cells × n_latent` batch-corrected embedding         |\n| `adata.obsm[\"X_scANVI\"]`       | label-aware embedding (better separates known classes) |\n| `adata.obs[\"pred_cell_type\"]`  | scANVI predicted label per cell                        |\n| `adata.layers[\"scvi_normalized\"]` | decoded expression, library-size normalized         |\n| DE dataframe                   | per-gene `lfc_*` \u002F `proba_de` (with `mode=\"change\"`)   |\n\n## Remote compute\n\nA100-class GPU recommended for >50k cells. The prebuilt **`singlecell_gpu`**\nModal env ships scvi-tools 1.4.2 + scanpy 1.11.5 + anndata 0.11.4 — read\n`compute_details({provider: 'byoc:modal', mode: 'read'})` for the current\nimage ref, then:\n\n```python\nc = host.compute.create('byoc:modal', provider_params={'modal': {\n    'image':  '\u003Cimage ref from compute_details>',   # e.g. im-...\n    'gpu':    'A100',\n    'cpu':    8,\n    'memory': 32768,\n    'timeout': 3600,\n}})\njob = c.submit_job(\n    intent=\"scVI+scANVI on 80k cells — 1×A100, ~15 min\",\n    inputs=[\n        {\"src\": \"dataset.h5ad\", \"dst_filename\": \"dataset.h5ad\"},\n        {\"src\": \"pipeline.py\",  \"dst_filename\": \"pipeline.py\"},\n    ],\n    command=\"python pipeline.py\",\n    outputs=[\"out\u002F**\"],\n    timeout_seconds=2400,\n)\nprint(job.job_id)   # cell ends here — kernel never blocks on compute\n```\n\n`h5ad_safe_obs` is auto-loaded into the **local** analysis kernel only — in\n`pipeline.py` running on Modal, paste the helper at the top of the script\n(or inline the `pd.Index(np.asarray(..., dtype=object))` coercion) before\n`.write_h5ad()`.\n\nThen call the `wait_for_notification` brain-tool. When `compute_done`\narrives, `save_artifacts(payload[\"featured_files\"])`. For the full result\ndict, re-enter the kernel and bind the **compute handle** (not the job)\nseparately — `.close()` lives on the handle, not on the job:\n\n```python\nh = host.compute.create('byoc:modal')\nres = h.attach_job(job_id).result()   # output_files, remote_workdir, ...\nh.close()\n```\n\nSee the `remote-compute-modal` skill for orchestration details.\n\n## Gotchas\n\n| Gotcha | What happens \u002F fix |\n|---|---|\n| `differential_expression()` defaults to `mode=\"vanilla\"` (scvi-tools ≥1.4) | `KeyError: 'lfc_mean'` \u002F `'proba_de'` when sorting — pass `mode=\"change\"` to get `lfc_*`\u002F`proba_de`\u002F`is_de_fdr_*`; in vanilla mode sort on `bayes_factor`. |\n| `adata.obs` index\u002Fcolumns are `string[pyarrow]` (`ArrowStringArray`) | `.write_h5ad()` dies with `IORegistryError: No method registered for writing \u003Cclass 'pandas.arrays.ArrowStringArray'>` (anndata #2377). Coerce before writing: `adata.obs = h5ad_safe_obs(adata.obs)` (kernel helper — local kernel only; inline the coercion in remote `pipeline.py`). **`.astype(str)` alone is not enough** — on a pyarrow-backed Index\u002FSeries it returns another Arrow-backed array; round-trip through `np.asarray(..., dtype=object)`. `anndata.settings.allow_write_nullable_strings = True` does **not** cover Arrow-backed strings. |\n| `use_gpu=` kwarg | Removed in 1.x → `TypeError: train() got an unexpected keyword argument 'use_gpu'`. Use `accelerator=\"gpu\", devices=1`. |\n| Log-normalized data fed to `setup_anndata` | Silent garbage — scVI's NB likelihood needs raw integer counts. Stash counts in `adata.layers[\"counts\"]` *before* normalize\u002Flog1p and pass `layer=\"counts\"`. |\n\n## Troubleshooting\n\n| Symptom | Fix |\n|---|---|\n| `KeyError: 'lfc_mean'` (or `'proba_de'`, `'is_de_fdr_0.05'`) on DE result | Add `mode=\"change\"` to `differential_expression()`; the default vanilla mode has no LFC columns. |\n| `IORegistryError: No method registered for writing \u003Cclass 'pandas.arrays.ArrowStringArray'>` on `.write_h5ad()` | `adata.obs = h5ad_safe_obs(adata.obs)` (and `adata.var` if needed) before writing. The `allow_write_nullable_strings` flag does not help here. |\n| `TypeError: ... unexpected keyword argument 'use_gpu'` | Replace with `accelerator=\"gpu\", devices=1`. |\n| `ValueError: ... non-negative integers` \u002F NB loss explodes | `layer=\"counts\"` points at log\u002Ffloat data — restore raw counts. |\n| `MisconfigurationException: No supported gpu backend found` | No CUDA visible — drop `accelerator`\u002F`devices` to fall back to CPU, or dispatch via Remote compute. |\n| `UnicodeEncodeError: 'ascii' codec can't encode character ...` writing a summary \u002F printing | Container has no `LANG` so Python defaults to ASCII. Open files with `encoding=\"utf-8\"` and\u002For `sys.stdout.reconfigure(encoding=\"utf-8\")` at script top. The prebuilt `singlecell_gpu` env sets `PYTHONIOENCODING=utf-8`, so this only bites user-built images. |\n| `AttributeError: ... object has no attribute 'close'` on a job handle | You chained `host.compute.create(...).attach_job(...)` and called `.close()` on the job. Bind the compute handle separately and close that — see Remote compute above. |\n\n---\n\n**Next**: cluster on `X_scVI` with scanpy (`sc.pp.neighbors(use_rep=\"X_scVI\")`\n→ `sc.tl.leiden` → `sc.tl.umap`); for spatial deconvolution train\ncell2location \u002F DestVI \u002F Tangram on the scRNA-seq reference.\n","# scvi-tools — scVI \u002F scANVI\n\nscvi-tools (Gayoso et al. 2022, github.com\u002Fscverse\u002Fscvi-tools, BSD-3-Clause)\nwraps a family\nof deep generative models for single-cell omics. The scRNA-seq core is **scVI**\n(unsupervised batch-corrected latent embedding) and **scANVI** (scVI + a\nclassifier head for semi-supervised cell-type label transfer). Both expect\n**raw integer UMI counts** and emit a low-dimensional `X_scVI` \u002F `X_scANVI`\nthat drops into the scanpy neighbors → leiden → umap pipeline.\n\n## How to run\n\n### scVI — batch-corrected latent space\n\n```python\nimport scanpy as sc\nimport scvi\n\nadata = sc.read_h5ad(\"dataset.h5ad\")\nadata.layers[\"counts\"] = adata.X.copy()          # preserve raw BEFORE any normalize\u002Flog1p\nsc.pp.normalize_total(adata); sc.pp.log1p(adata) # optional, for HVG \u002F plotting only\nsc.pp.highly_variable_genes(adata, n_top_genes=2000, batch_key=\"batch\", subset=True)\n\nscvi.model.SCVI.setup_anndata(adata, layer=\"counts\", batch_key=\"batch\")\nmodel = scvi.model.SCVI(adata, n_latent=30)\nmodel.train(max_epochs=200, early_stopping=True, accelerator=\"gpu\", devices=1)\n\nadata.obsm[\"X_scVI\"] = model.get_latent_representation()\nadata.layers[\"scvi_normalized\"] = model.get_normalized_expression(library_size=1e4)\n```\n\n### scANVI — label transfer from a partially-annotated reference\n\n```python\nlvae = scvi.model.SCANVI.from_scvi_model(\n    model, labels_key=\"cell_type\", unlabeled_category=\"Unknown\",\n)\nlvae.train(max_epochs=20, n_samples_per_label=100, accelerator=\"gpu\", devices=1)\n\nadata.obsm[\"X_scANVI\"] = lvae.get_latent_representation()\nadata.obs[\"pred_cell_type\"] = lvae.predict()\n```\n\n`accelerator=\"gpu\", devices=1` is the PyTorch-Lightning spelling; the legacy\n`use_gpu=` kwarg was **removed** in scvi-tools 1.x and now raises `TypeError`.\n\n## Differential expression\n\n```python\nde = model.differential_expression(\n    groupby=\"leiden\", group1=\"3\",   # group2=None → vs. all other cells\n    mode=\"change\", delta=0.25,\n)\ntop = de.sort_values(\"proba_de\", ascending=False).head(50)\n```\n\nFor one-vs-rest leave `group2` out — `\"rest\"` is scanpy's\n`rank_genes_groups` convention, not scvi-tools'; here `group2` is a literal\ncategory name and `\"rest\"` would match zero cells.\n\nscvi-tools ≥1.4 defaults to `mode=\"vanilla\"`, whose result columns are\nexactly:\n\n```\n['proba_m1', 'proba_m2', 'bayes_factor', 'scale1', 'scale2', 'raw_mean1',\n 'raw_mean2', 'non_zeros_proportion1', 'non_zeros_proportion2',\n 'raw_normalized_mean1', 'raw_normalized_mean2', 'comparison', 'group1',\n 'group2']\n```\n\n— no `lfc_*`, no `proba_de`, no `is_de_fdr_*`. **Pass `mode=\"change\"`** to\nget `lfc_mean` \u002F `lfc_median` \u002F `proba_de` \u002F `is_de_fdr_0.05`. Sort on\n`proba_de` (or on `bayes_factor` if you deliberately stayed in vanilla\nmode).\n\n## Output format\n\n| Key                            | What                                                   |\n| ------------------------------ | ------------------------------------------------------ |\n| `adata.obsm[\"X_scVI\"]`         | `n_cells × n_latent` batch-corrected embedding         |\n| `adata.obsm[\"X_scANVI\"]`       | label-aware embedding (better separates known classes) |\n| `adata.obs[\"pred_cell_type\"]`  | scANVI predicted label per cell                        |\n| `adata.layers[\"scvi_normalized\"]` | decoded expression, library-size normalized         |\n| DE dataframe                   | per-gene `lfc_*` \u002F `proba_de` (with `mode=\"change\"`)   |\n\n## Remote compute\n\nA100-class GPU recommended for >50k cells. The prebuilt **`singlecell_gpu`**\nModal env ships scvi-tools 1.4.2 + scanpy 1.11.5 + anndata 0.11.4 — read\n`compute_details({provider: 'byoc:modal', mode: 'read'})` for the current\nimage ref, then:\n\n```python\nc = host.compute.create('byoc:modal', provider_params={'modal': {\n    'image':  '\u003Cimage ref from compute_details>',   # e.g. im-...\n    'gpu':    'A100',\n    'cpu':    8,\n    'memory': 32768,\n    'timeout': 3600,\n}})\njob = c.submit_job(\n    intent=\"scVI+scANVI on 80k cells — 1×A100, ~15 min\",\n    inputs=[\n        {\"src\": \"dataset.h5ad\", \"dst_filename\": \"dataset.h5ad\"},\n        {\"src\": \"pipeline.py\",  \"dst_filename\": \"pipeline.py\"},\n    ],\n    command=\"python pipeline.py\",\n    outputs=[\"out\u002F**\"],\n    timeout_seconds=2400,\n)\nprint(job.job_id)   # cell ends here — kernel never blocks on compute\n```\n\n`h5ad_safe_obs` is auto-loaded into the **local** analysis kernel only — in\n`pipeline.py` running on Modal, paste the helper at the top of the script\n(or inline the `pd.Index(np.asarray(..., dtype=object))` coercion) before\n`.write_h5ad()`.\n\nThen call the `wait_for_notification` brain-tool. When `compute_done`\narrives, `save_artifacts(payload[\"featured_files\"])`. For the full result\ndict, re-enter the kernel and bind the **compute handle** (not the job)\nseparately — `.close()` lives on the handle, not on the job:\n\n```python\nh = host.compute.create('byoc:modal')\nres = h.attach_job(job_id).result()   # output_files, remote_workdir, ...\nh.close()\n```\n\nSee the `remote-compute-modal` skill for orchestration details.\n\n## Gotchas\n\n| Gotcha | What happens \u002F fix |\n|---|---|\n| `differential_expression()` defaults to `mode=\"vanilla\"` (scvi-tools ≥1.4) | `KeyError: 'lfc_mean'` \u002F `'proba_de'` when sorting — pass `mode=\"change\"` to get `lfc_*`\u002F`proba_de`\u002F`is_de_fdr_*`; in vanilla mode sort on `bayes_factor`. |\n| `adata.obs` index\u002Fcolumns are `string[pyarrow]` (`ArrowStringArray`) | `.write_h5ad()` dies with `IORegistryError: No method registered for writing \u003Cclass 'pandas.arrays.ArrowStringArray'>` (anndata #2377). Coerce before writing: `adata.obs = h5ad_safe_obs(adata.obs)` (kernel helper — local kernel only; inline the coercion in remote `pipeline.py`). **`.astype(str)` alone is not enough** — on a pyarrow-backed Index\u002FSeries it returns another Arrow-backed array; round-trip through `np.asarray(..., dtype=object)`. `anndata.settings.allow_write_nullable_strings = True` does **not** cover Arrow-backed strings. |\n| `use_gpu=` kwarg | Removed in 1.x → `TypeError: train() got an unexpected keyword argument 'use_gpu'`. Use `accelerator=\"gpu\", devices=1`. |\n| Log-normalized data fed to `setup_anndata` | Silent garbage — scVI's NB likelihood needs raw integer counts. Stash counts in `adata.layers[\"counts\"]` *before* normalize\u002Flog1p and pass `layer=\"counts\"`. |\n\n## Troubleshooting\n\n| Symptom | Fix |\n|---|---|\n| `KeyError: 'lfc_mean'` (or `'proba_de'`, `'is_de_fdr_0.05'`) on DE result | Add `mode=\"change\"` to `differential_expression()`; the default vanilla mode has no LFC columns. |\n| `IORegistryError: No method registered for writing \u003Cclass 'pandas.arrays.ArrowStringArray'>` on `.write_h5ad()` | `adata.obs = h5ad_safe_obs(adata.obs)` (and `adata.var` if needed) before writing. The `allow_write_nullable_strings` flag does not help here. |\n| `TypeError: ... unexpected keyword argument 'use_gpu'` | Replace with `accelerator=\"gpu\", devices=1`. |\n| `ValueError: ... non-negative integers` \u002F NB loss explodes | `layer=\"counts\"` points at log\u002Ffloat data — restore raw counts. |\n| `MisconfigurationException: No supported gpu backend found` | No CUDA visible — drop `accelerator`\u002F`devices` to fall back to CPU, or dispatch via Remote compute. |\n| `UnicodeEncodeError: 'ascii' codec can't encode character ...` writing a summary \u002F printing | Container has no `LANG` so Python defaults to ASCII. Open files with `encoding=\"utf-8\"` and\u002For `sys.stdout.reconfigure(encoding=\"utf-8\")` at script top. The prebuilt `singlecell_gpu` env sets `PYTHONIOENCODING=utf-8`, so this only bites user-built images. |\n| `AttributeError: ... object has no attribute 'close'` on a job handle | You chained `host.compute.create(...).attach_job(...)` and called `.close()` on the job. Bind the compute handle separately and close that — see Remote compute above. |\n\n---\n\n**Next**: cluster on `X_scVI` with scanpy (`sc.pp.neighbors(use_rep=\"X_scVI\")`\n→ `sc.tl.leiden` → `sc.tl.umap`); for spatial deconvolution train\ncell2location \u002F DestVI \u002F Tangram on the scRNA-seq reference.\n",{"name":5,"license":15,"metadata":24,"description":25,"requirements":26},{"display-name":5},"Probabilistic single-cell RNA-seq with scvi-tools — scVI for a batch-corrected latent space, scANVI for semi-supervised label transfer, and Bayesian differential expression. Reach for this skill to integrate scRNA-seq batches, embed cells for clustering, transfer annotations from a reference onto a query, or score differentially expressed genes per cluster. For spatial deconvolution \u002F mapping use the cell2location, DestVI, or Tangram methods instead.\n",[27],"gpu","main","e71245365df78e06cd166602744bf2491f79910e","research-skills-hub\u002Fclaude-science-skills\u002Fskills\u002Fscvi-tools",0,"https:\u002F\u002Fgithub.com\u002Fpengqianhan\u002FAI-Human-Research-OS","https:\u002F\u002Fpengqianhan.github.io\u002FAI-Human-Research-OS\u002F","https:\u002F\u002Fgithub.com\u002Fpengqianhan","Ph.D. candidate of The University of Auckland, New Zealand","2026-11-14 18:08:16",{"source":8,"skillName":5,"commands":38,"prompt":48,"zipAvailable":49},[39,42,45],{"tool":40,"command":41},"npx","npx skills add pengqianhan\u002FAI-Human-Research-OS --skill scvi-tools",{"tool":43,"command":44},"bunx","bunx skills add pengqianhan\u002FAI-Human-Research-OS --skill scvi-tools",{"tool":46,"command":47},"pnpm","pnpm dlx skills add pengqianhan\u002FAI-Human-Research-OS --skill scvi-tools","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fpengqianhan\u002FAI-Human-Research-OS\u002Fblob\u002Fmain\u002Fresearch-skills-hub\u002Fclaude-science-skills\u002Fskills\u002Fscvi-tools\u002FSKILL.md",true,[51],{"path":52,"resourceType":53,"resourceTypeDesc":54,"sizeBytes":55,"sourceHtmlUrl":56},"kernel.py","SCRIPT","脚本",2248,"https:\u002F\u002Fgithub.com\u002Fpengqianhan\u002FAI-Human-Research-OS\u002Fblob\u002Fmain\u002Fresearch-skills-hub\u002Fclaude-science-skills\u002Fskills\u002Fscvi-tools\u002Fkernel.py",[58,66,73,80,87,94],{"skillNo":59,"name":60,"nameZh":6,"description":61,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":62,"licenseFileUrl":6,"sourceHtmlUrl":63,"categories":64,"keywords":65,"publishedAt":19},"01M74MWYTHT498SVMN823896T7","workflow-skill-creator","Distills a completed user workflow or interaction into a reusable agent skill. Use when the user asks to turn their workflow, interaction, or multi-step process into a skill, or when they say \"make this a skill\", \"create a skill from what we just did\", \"package this workflow\" or similar. Do not use for creating skills from scratch without an existing workflow (use a generic skill-creator for that).","MIT","https:\u002F\u002Fgithub.com\u002Fpengqianhan\u002FAI-Human-Research-OS\u002Fblob\u002Fmain\u002Fresearch-skills-hub\u002Fscience-skills\u002Fworkflow_skill_creator\u002FSKILL.md",[],[],{"skillNo":67,"name":68,"nameZh":6,"description":69,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":62,"licenseFileUrl":6,"sourceHtmlUrl":70,"categories":71,"keywords":72,"publishedAt":19},"01M74MWWVCFZ8HQS8840F1V1FT","uv","Checks whether the uv Python package manager is installed and installs it if missing. Ensures uv is on PATH. Use when another skill requires uv as a prerequisite.","https:\u002F\u002Fgithub.com\u002Fpengqianhan\u002FAI-Human-Research-OS\u002Fblob\u002Fmain\u002Fresearch-skills-hub\u002Fscience-skills\u002Fuv\u002FSKILL.md",[],[],{"skillNo":74,"name":75,"nameZh":6,"description":76,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":62,"licenseFileUrl":6,"sourceHtmlUrl":77,"categories":78,"keywords":79,"publishedAt":19},"01M74MWRH5BTSWZ2C5PCTD5FV9","uniprot-database","Access protein metadata, function, taxonomy, and sequences across UniProtKB, UniParc, and UniRef. Use when searching for proteins, mapping identifiers, or retrieving functional annotations and publications. Don't use for sequence alignment, protein folding, or sequence similarity search (use specialized skills for those tasks).","https:\u002F\u002Fgithub.com\u002Fpengqianhan\u002FAI-Human-Research-OS\u002Fblob\u002Fmain\u002Fresearch-skills-hub\u002Fscience-skills\u002Funiprot_database\u002FSKILL.md",[],[],{"skillNo":81,"name":82,"nameZh":6,"description":83,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":62,"licenseFileUrl":6,"sourceHtmlUrl":84,"categories":85,"keywords":86,"publishedAt":19},"01M74MWN1D3ER9F1FZPWH8HQC1","unibind-database","Queries the UniBind database for experimentally validated transcription factor (TF) binding sites. Use when retrieving direct TF-DNA interaction datasets, downloading binding site coordinates (BED\u002FFASTA) for local analysis, or listing available datasets by species, cell line, or TF name. Don't use to query specific intervals, locations, genes, motif models or expression data.","https:\u002F\u002Fgithub.com\u002Fpengqianhan\u002FAI-Human-Research-OS\u002Fblob\u002Fmain\u002Fresearch-skills-hub\u002Fscience-skills\u002Funibind_database\u002FSKILL.md",[],[],{"skillNo":88,"name":89,"nameZh":6,"description":90,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":62,"licenseFileUrl":6,"sourceHtmlUrl":91,"categories":92,"keywords":93,"publishedAt":19},"01M74MWHHB5QZY3NR2FYA1EK4N","ucsc-conservation-and-tfbs","Fetch Evolutionary Conservation scores (phyloP, phastCons) and Transcription Factor Binding Sites (TFBS) from the UCSC Genome Browser. Use when analyzing whether genomic variants or regions are evolutionarily conserved, functionally important, or bounded by TF regulators across major projects (ENCODE, JASPAR, ReMap).","https:\u002F\u002Fgithub.com\u002Fpengqianhan\u002FAI-Human-Research-OS\u002Fblob\u002Fmain\u002Fresearch-skills-hub\u002Fscience-skills\u002Fucsc_conservation_and_tfbs\u002FSKILL.md",[],[],{"skillNo":95,"name":96,"nameZh":6,"description":97,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":62,"licenseFileUrl":6,"sourceHtmlUrl":98,"categories":99,"keywords":100,"publishedAt":19},"01M74MWERYHSY87P21FDMVE6BA","string-database","Query the STRING database for protein-protein interactions (PPIs), functional enrichment, and homology. Use when the user asks about interactions between specific proteins, interaction evidence, confidence scores, protein interaction partners, or pathway enrichments.","https:\u002F\u002Fgithub.com\u002Fpengqianhan\u002FAI-Human-Research-OS\u002Fblob\u002Fmain\u002Fresearch-skills-hub\u002Fscience-skills\u002Fstring_database\u002FSKILL.md",[],[]]