@wentorai/research-plugins 1.3.1 → 1.4.0

package/skills/domains/ai-ml/huggingface-api/SKILL.md

@@ -0,0 +1,251 @@
+---
+name: huggingface-api
+description: "Search and discover ML models, datasets, and Spaces on Hugging Face"
+metadata:
+  openclaw:
+    emoji: "🤗"
+    category: "domains"
+    subcategory: "ai-ml"
+    keywords: ["Hugging Face", "ML models", "datasets", "model hub", "transformers", "NLP", "computer vision"]
+    source: "https://huggingface.co/docs/hub/api"
+---
+
+# Hugging Face Hub API
+
+## Overview
+
+The Hugging Face Hub is the largest open-source ML ecosystem, hosting over 1 million models, 200,000+ datasets, and 400,000+ Spaces (demo apps). The Hub API at `https://huggingface.co/api` provides programmatic access to search, discover, and retrieve metadata for all public resources without authentication.
+
+For academic researchers, the Hub API enables systematic model selection for benchmarking, dataset discovery for experiments, tracking community adoption metrics (downloads, likes), and building reproducible ML pipelines that reference specific model revisions by SHA.
+
+## Authentication
+
+**Read endpoints require no authentication.** All search and metadata queries work without a token.
+
+For write operations (uploading models, creating repos), set a User Access Token:
+
+```bash
+export HF_TOKEN="hf_..."
+# Pass via header:
+curl -H "Authorization: Bearer $HF_TOKEN" https://huggingface.co/api/...
+```
+
+Generate tokens at: https://huggingface.co/settings/tokens
+
+## Core Endpoints
+
+### Search Models
+
+```
+GET https://huggingface.co/api/models?search={query}&limit={n}&sort={field}&direction={-1|1}
+```
+
+**Parameters**: `search` (query string), `limit` (max results), `sort` (field: `downloads`, `likes`, `lastModified`, `trending`), `direction` (-1 descending, 1 ascending), `filter` (pipeline tag like `text-classification`), `author` (org/user filter), `library` (e.g. `transformers`, `pytorch`)
+
+**Example** -- top 2 models for "bert" by downloads:
+
+```bash
+curl -s "https://huggingface.co/api/models?search=bert&limit=2&sort=downloads&direction=-1"
+```
+
+```json
+[
+  {
+    "id": "google-bert/bert-base-uncased",
+    "likes": 2587,
+    "downloads": 71053483,
+    "pipeline_tag": "fill-mask",
+    "library_name": "transformers",
+    "tags": ["transformers","pytorch","tf","jax","bert","fill-mask","en",
+             "dataset:bookcorpus","dataset:wikipedia","arxiv:1810.04805",
+             "license:apache-2.0"]
+  },
+  {
+    "id": "google-bert/bert-base-multilingual-uncased",
+    "likes": 153,
+    "downloads": 5017183,
+    "pipeline_tag": "fill-mask",
+    "library_name": "transformers"
+  }
+]
+```
+
+### Get Model Details
+
+```
+GET https://huggingface.co/api/models/{owner}/{model_name}
+```
+
+Returns full metadata including `config.architectures`, `cardData` (license, datasets, language), `siblings` (file listing), `sha` (exact revision), and `lastModified`.
+
+```bash
+curl -s "https://huggingface.co/api/models/google-bert/bert-base-uncased"
+```
+
+Key fields in response:
+
+```json
+{
+  "id": "google-bert/bert-base-uncased",
+  "sha": "86b5e0934494bd15c9632b12f734a8a67f723594",
+  "lastModified": "2024-02-19T11:06:12.000Z",
+  "downloads": 71053483,
+  "config": { "architectures": ["BertForMaskedLM"], "model_type": "bert" },
+  "cardData": { "language": "en", "license": "apache-2.0",
+                "datasets": ["bookcorpus","wikipedia"] }
+}
+```
+
+### Search Datasets
+
+```
+GET https://huggingface.co/api/datasets?search={query}&limit={n}
+```
+
+**Parameters**: `search`, `limit`, `sort`, `direction`, `author`, `filter` (task tag like `question-answering`)
+
+```bash
+curl -s "https://huggingface.co/api/datasets?search=squad&limit=2"
+```
+
+```json
+[
+  {
+    "id": "rajpurkar/squad_v2",
+    "likes": 242,
+    "downloads": 36017,
+    "description": "Stanford Question Answering Dataset (SQuAD)...",
+    "tags": ["task_categories:question-answering","language:en",
+             "license:cc-by-sa-4.0","size_categories:100K<n<1M",
+             "arxiv:1806.03822"]
+  }
+]
+```
+
+### Get Dataset Details
+
+```
+GET https://huggingface.co/api/datasets/{owner}/{dataset_name}
+```
+
+```bash
+curl -s "https://huggingface.co/api/datasets/rajpurkar/squad_v2"
+```
+
+Returns `cardData` with structured metadata (task categories, languages, license, size), `description`, `paperswithcode_id` for cross-referencing, and `tags` with arXiv paper IDs.
+
+### Search Spaces
+
+```
+GET https://huggingface.co/api/spaces?search={query}&limit={n}
+```
+
+```bash
+curl -s "https://huggingface.co/api/spaces?search=chatbot&limit=2"
+```
+
+```json
+[
+  {
+    "id": "21Hg/chatbot",
+    "likes": 5,
+    "sdk": "docker",
+    "tags": ["docker","streamlit","region:us"]
+  },
+  {
+    "id": "lmarena-ai/chatbot-arena",
+    "likes": 234,
+    "sdk": "static"
+  }
+]
+```
+
+## Advanced Filters
+
+Combine filters via query params to narrow results:
+
+```bash
+# PyTorch text-generation models with 1000+ likes
+curl -s "https://huggingface.co/api/models?filter=text-generation&library=pytorch&sort=likes&direction=-1&limit=5"
+
+# Datasets for NER tasks in Chinese
+curl -s "https://huggingface.co/api/datasets?filter=token-classification&language=zh&limit=10"
+
+# Gradio Spaces sorted by trending
+curl -s "https://huggingface.co/api/spaces?filter=gradio&sort=trending&direction=-1&limit=5"
+```
+
+## Rate Limits
+
+- **Unauthenticated**: generous but undocumented; suitable for interactive use and small scripts
+- **Authenticated**: higher limits with Bearer token
+- **Best practice**: add `limit` parameter to avoid fetching thousands of results; cache responses locally for batch analysis
+- No strict per-minute quota is published; if you receive HTTP 429, back off exponentially
+
+## Academic Use Cases
+
+1. **Model selection for benchmarks**: Search by pipeline tag (`text-classification`, `token-classification`, `summarization`) and sort by downloads to find community-validated baselines
+2. **Dataset discovery**: Filter by `task_categories`, `language`, and `size_categories` tags to find training data matching your experimental requirements
+3. **Reproducibility**: Pin model versions using the `sha` field from model details -- load exact revisions with `revision="86b5e093..."` in transformers
+4. **Citation tracking**: Extract `arxiv:` tags from model/dataset metadata to trace foundational papers
+5. **Ecosystem analysis**: Aggregate download/like counts across model families to study adoption trends in ML research
+
+## Code Examples
+
+### Python with requests
+
+```python
+import requests
+
+# Search for top text-classification models
+resp = requests.get("https://huggingface.co/api/models", params={
+    "filter": "text-classification",
+    "sort": "downloads",
+    "direction": -1,
+    "limit": 10
+})
+models = resp.json()
+for m in models:
+    print(f"{m['id']:50s}  downloads={m.get('downloads',0):>12,}")
+
+# Get specific model metadata
+detail = requests.get("https://huggingface.co/api/models/google-bert/bert-base-uncased").json()
+print(f"SHA: {detail['sha']}")
+print(f"License: {detail['cardData'].get('license')}")
+```
+
+### Python with huggingface_hub library
+
+```python
+from huggingface_hub import HfApi
+
+api = HfApi()
+
+# Search models (returns ModelInfo objects)
+models = api.list_models(search="bert", sort="downloads", direction=-1, limit=5)
+for m in models:
+    print(f"{m.id}  downloads={m.downloads}")
+
+# Get full model info
+info = api.model_info("google-bert/bert-base-uncased")
+print(f"Pipeline: {info.pipeline_tag}, SHA: {info.sha}")
+
+# Search datasets
+datasets = api.list_datasets(search="squad", sort="downloads", direction=-1, limit=5)
+for d in datasets:
+    print(f"{d.id}  downloads={d.downloads}")
+
+# List Spaces
+spaces = api.list_spaces(search="chatbot", limit=5)
+for s in spaces:
+    print(f"{s.id}  sdk={s.sdk}")
+```
+
+## References
+
+- Hub API documentation: https://huggingface.co/docs/hub/api
+- huggingface_hub Python library: https://huggingface.co/docs/huggingface_hub/
+- Model Hub: https://huggingface.co/models
+- Dataset Hub: https://huggingface.co/datasets
+- Spaces: https://huggingface.co/spaces
+- OpenAPI spec: https://huggingface.co/docs/hub/api#openapi

package/skills/domains/ai-ml/llm-evaluation-guide/SKILL.md

@@ -3,7 +3,7 @@ name: llm-evaluation-guide
 description: "Evaluate and benchmark large language models for research applications"
 metadata:
   openclaw:
-    emoji: "brain"
+    emoji: "🧠"
     category: "domains"
     subcategory: "ai-ml"
     keywords: ["LLM evaluation", "benchmarking", "language models", "model evaluation", "NLP metrics", "BLEU", "perplexity"]

package/skills/domains/ai-ml/prompt-engineering-research/SKILL.md

@@ -3,7 +3,7 @@ name: prompt-engineering-research
 description: "Systematic prompt engineering methods for AI-assisted academic research workf..."
 metadata:
   openclaw:
-    emoji: "robot"
+    emoji: "🤖"
     category: "domains"
     subcategory: "ai-ml"
     keywords: ["machine learning", "deep learning", "NLP", "AI coding", "prompt engineering", "LLM"]

package/skills/domains/ai-ml/reinforcement-learning-guide/SKILL.md

@@ -3,7 +3,7 @@ name: reinforcement-learning-guide
 description: "Reinforcement learning fundamentals, algorithms, and research"
 metadata:
   openclaw:
-    emoji: "robot"
+    emoji: "🤖"
     category: "domains"
     subcategory: "ai-ml"
     keywords: ["reinforcement learning", "machine learning", "deep learning", "neural network"]

package/skills/domains/ai-ml/transformer-architecture-guide/SKILL.md

@@ -3,7 +3,7 @@ name: transformer-architecture-guide
 description: "Guide to Transformer architectures for NLP and computer vision"
 metadata:
   openclaw:
-    emoji: "brain"
+    emoji: "🧠"
     category: "domains"
     subcategory: "ai-ml"
     keywords: ["Transformer", "neural network", "deep learning", "NLP", "computer vision"]

package/skills/domains/biomedical/SKILL.md

@@ -1,20 +1,25 @@
 ---
 name: biomedical-skills
-description: "17 biomedical research skills (bioagents, clawbio, clinical-dialogue-agents, clinical-research...). Trigger: medical research, clinical trials, genomics, bioinformatics. Design: domain databases, wet-lab/dry-lab methods, and ethical compliance guides."
+description: "24 biomedical research skills. Trigger: medical research, clinical trials, genomics, bioinformatics. Design: domain databases, wet-lab/dry-lab methods, and ethical compliance guides."
 ---
 
-# Biomedical Research — 17 Skills
+# Biomedical Research — 24 Skills
 
 Select the skill matching the user's need, then `read` its SKILL.md.
 
 | Skill | Description |
 |-------|-------------|
+| [alphafold-api](./alphafold-api/SKILL.md) | Query AlphaFold protein structure predictions by UniProt accession |
 | [bioagents-guide](./bioagents-guide/SKILL.md) | AI scientist framework for autonomous biological research workflows |
+| [biothings-api](./biothings-api/SKILL.md) | Query gene, variant, and drug annotations via BioThings APIs |
 | [clawbio-guide](./clawbio-guide/SKILL.md) | OpenClaw bioinformatics skill library for genomics pipelines |
 | [clinical-dialogue-agents-guide](./clinical-dialogue-agents-guide/SKILL.md) | Papers on AI agents for clinical dialogue and medical QA |
 | [clinical-research-guide](./clinical-research-guide/SKILL.md) | Design clinical studies and report using CONSORT, STROBE guidelines |
 | [clinicaltrials-api](./clinicaltrials-api/SKILL.md) | Clinical trial registry database search API |
+| [clinicaltrials-api-v2](./clinicaltrials-api-v2/SKILL.md) | Search and analyze clinical trials via the ClinicalTrials.gov v2 API |
 | [ena-sequence-api](./ena-sequence-api/SKILL.md) | Access nucleotide sequence data from the European Nucleotide Archive |
+| [enrichr-api](./enrichr-api/SKILL.md) | Perform gene set enrichment analysis using the Enrichr API |
+| [ensembl-rest-api](./ensembl-rest-api/SKILL.md) | Query gene, sequence, and variant data via the Ensembl REST API |
 | [epidemiology-guide](./epidemiology-guide/SKILL.md) | Epidemiological study designs, measures of association, and public health ana... |
 | [genomas-guide](./genomas-guide/SKILL.md) | Automate gene expression analysis with the GenoMAS multi-agent system |
 | [genomics-analysis-guide](./genomics-analysis-guide/SKILL.md) | Workflows for RNA-seq, GWAS, and variant calling in genomic research |
@@ -22,7 +27,9 @@ Select the skill matching the user's need, then `read` its SKILL.md.
 | [med-researcher-guide](./med-researcher-guide/SKILL.md) | Multi-agent system for biomedical literature review and synthesis |
 | [med-researcher-r1-guide](./med-researcher-r1-guide/SKILL.md) | Medical deep research agent with reasoning chain analysis |
 | [medgeclaw-guide](./medgeclaw-guide/SKILL.md) | AI research assistant for biomedicine, RNA-seq, and drug discovery |
+| [medical-data-api](./medical-data-api/SKILL.md) | Access FDA drug data and WHO global health statistics for research |
 | [medical-imaging-guide](./medical-imaging-guide/SKILL.md) | Medical image analysis with deep learning for research applications |
 | [ncbi-blast-api](./ncbi-blast-api/SKILL.md) | Run sequence similarity searches via the NCBI BLAST REST API |
 | [ncbi-datasets-api](./ncbi-datasets-api/SKILL.md) | Access genomes, genes, and taxonomy data via NCBI Datasets v2 API |
+| [pdb-structure-api](./pdb-structure-api/SKILL.md) | Search and retrieve 3D protein structures from the RCSB Protein Data Bank |
 | [quickgo-api](./quickgo-api/SKILL.md) | Browse and search Gene Ontology annotations via the QuickGO API |

package/skills/domains/biomedical/alphafold-api/SKILL.md

@@ -0,0 +1,227 @@
+---
+name: alphafold-api
+description: "Query AlphaFold protein structure predictions by UniProt accession"
+metadata:
+  openclaw:
+    emoji: "🧬"
+    category: "domains"
+    subcategory: "biomedical"
+    keywords: ["protein structure", "AlphaFold", "structure prediction", "UniProt", "pLDDT", "bioinformatics"]
+    source: "https://alphafold.ebi.ac.uk"
+---
+
+# AlphaFold Protein Structure Database API
+
+## Overview
+
+The AlphaFold DB, maintained by EMBL-EBI and DeepMind, provides open access to over 200 million protein structure predictions. The REST API enables programmatic lookup of predicted structures, confidence metrics (pLDDT, PAE), and downloadable structure files (PDB, mmCIF, BinaryCIF) keyed on UniProt accessions. Free, no authentication required.
+
+## Authentication
+
+None. All endpoints are publicly accessible without API keys or tokens.
+
+## Core Endpoints
+
+Base URL: `https://alphafold.ebi.ac.uk/api`
+
+### 1. Get Prediction by UniProt Accession
+
+Retrieves all AlphaFold models for a given UniProt accession or model ID.
+
+```bash
+curl "https://alphafold.ebi.ac.uk/api/prediction/P04637"
+```
+
+**Response** (first entry, abbreviated):
+
+```json
+[
+  {
+    "entryId": "AF-P04637-F1",
+    "uniprotAccession": "P04637",
+    "uniprotId": "P53_HUMAN",
+    "uniprotDescription": "Cellular tumor antigen p53",
+    "gene": "TP53",
+    "organismScientificName": "Homo sapiens",
+    "taxId": 9606,
+    "globalMetricValue": 75.06,
+    "fractionPlddtVeryHigh": 0.527,
+    "fractionPlddtConfident": 0.071,
+    "fractionPlddtLow": 0.104,
+    "fractionPlddtVeryLow": 0.298,
+    "latestVersion": 6,
+    "modelCreatedDate": "2025-08-01T00:00:00Z",
+    "sequenceStart": 1,
+    "sequenceEnd": 393,
+    "pdbUrl": "https://alphafold.ebi.ac.uk/files/AF-P04637-F1-model_v6.pdb",
+    "cifUrl": "https://alphafold.ebi.ac.uk/files/AF-P04637-F1-model_v6.cif",
+    "bcifUrl": "https://alphafold.ebi.ac.uk/files/AF-P04637-F1-model_v6.bcif",
+    "paeImageUrl": "https://alphafold.ebi.ac.uk/files/AF-P04637-F1-predicted_aligned_error_v6.png",
+    "paeDocUrl": "https://alphafold.ebi.ac.uk/files/AF-P04637-F1-predicted_aligned_error_v6.json",
+    "plddtDocUrl": "https://alphafold.ebi.ac.uk/files/AF-P04637-F1-confidence_v6.json",
+    "amAnnotationsUrl": "https://alphafold.ebi.ac.uk/files/AF-P04637-F1-aa-substitutions.csv"
+  }
+]
+```
+
+### 2. Per-Residue Confidence Scores (pLDDT)
+
+Download the per-residue pLDDT confidence JSON linked in `plddtDocUrl`:
+
+```bash
+curl "https://alphafold.ebi.ac.uk/files/AF-P04637-F1-confidence_v6.json"
+```
+
+**Response** (truncated):
+
+```json
+{
+  "residueNumber": [1, 2, 3, 4, 5],
+  "confidenceScore": [40.66, 44.53, 49.97, 48.59, 44.88],
+  "confidenceCategory": ["D", "D", "D", "D", "D"]
+}
+```
+
+Categories: **A** (Very High, >90), **B** (Confident, 70-90), **C** (Low, 50-70), **D** (Very Low, <50).
+
+### 3. UniProt Summary (3D-Beacons Format)
+
+Returns model metadata following the 3D-Beacons data standard:
+
+```bash
+curl "https://alphafold.ebi.ac.uk/api/uniprot/summary/P04637.json"
+```
+
+**Response** (abbreviated):
+
+```json
+{
+  "uniprot_entry": {
+    "ac": "P04637",
+    "id": "P53_HUMAN",
+    "sequence_length": 393
+  },
+  "structures": [
+    {
+      "summary": {
+        "model_identifier": "AF-P04637-F1",
+        "model_url": "https://alphafold.ebi.ac.uk/files/AF-P04637-F1-model_v6.cif",
+        "provider": "AlphaFold DB",
+        "confidence_type": "pLDDT",
+        "confidence_avg_local_score": 75.06,
+        "coverage": 1.0
+      }
+    }
+  ]
+}
+```
+
+### 4. Download Structure Files
+
+Structure files are available at the URLs returned in prediction responses:
+
+```bash
+# PDB format
+curl -O "https://alphafold.ebi.ac.uk/files/AF-P04637-F1-model_v6.pdb"
+
+# mmCIF format
+curl -O "https://alphafold.ebi.ac.uk/files/AF-P04637-F1-model_v6.cif"
+
+# Predicted Aligned Error (PAE) matrix
+curl -O "https://alphafold.ebi.ac.uk/files/AF-P04637-F1-predicted_aligned_error_v6.json"
+```
+
+## Key Response Fields
+
+| Field | Type | Description |
+|-------|------|-------------|
+| `entryId` | string | AlphaFold model ID (e.g., `AF-P04637-F1`) |
+| `uniprotAccession` | string | UniProt accession code |
+| `gene` | string | Gene symbol |
+| `globalMetricValue` | float | Average pLDDT score (0-100) |
+| `fractionPlddtVeryHigh` | float | Fraction of residues with pLDDT > 90 |
+| `fractionPlddtConfident` | float | Fraction with pLDDT 70-90 |
+| `fractionPlddtLow` | float | Fraction with pLDDT 50-70 |
+| `fractionPlddtVeryLow` | float | Fraction with pLDDT < 50 |
+| `pdbUrl` | string | Direct download URL for PDB file |
+| `cifUrl` | string | Direct download URL for mmCIF file |
+| `paeDocUrl` | string | URL for predicted aligned error JSON |
+| `plddtDocUrl` | string | URL for per-residue confidence JSON |
+| `latestVersion` | int | Model version number |
+
+## Rate Limits
+
+The AlphaFold DB API has no published per-request rate limits. EMBL-EBI's general fair use policy applies: usage that degrades service for others may result in blocking. For bulk downloads (entire proteomes), use the FTP archive at `https://ftp.ebi.ac.uk/pub/databases/alphafold/` rather than repeated API calls.
+
+## Python Example
+
+```python
+import requests
+
+
+def get_alphafold_prediction(uniprot_id: str) -> dict:
+    """Fetch AlphaFold structure prediction for a UniProt accession."""
+    url = f"https://alphafold.ebi.ac.uk/api/prediction/{uniprot_id}"
+    resp = requests.get(url)
+    resp.raise_for_status()
+    entries = resp.json()
+    # Return the canonical (first) entry
+    return entries[0] if entries else None
+
+
+def get_confidence_scores(prediction: dict) -> dict:
+    """Download per-residue pLDDT confidence scores."""
+    resp = requests.get(prediction["plddtDocUrl"])
+    resp.raise_for_status()
+    return resp.json()
+
+
+def download_structure(prediction: dict, fmt: str = "pdb",
+                       output_dir: str = ".") -> str:
+    """Download structure file in pdb, cif, or bcif format."""
+    url_key = {"pdb": "pdbUrl", "cif": "cifUrl", "bcif": "bcifUrl"}[fmt]
+    url = prediction[url_key]
+    filename = url.split("/")[-1]
+    path = f"{output_dir}/{filename}"
+
+    resp = requests.get(url)
+    resp.raise_for_status()
+    with open(path, "wb") as f:
+        f.write(resp.content)
+    return path
+
+
+# Example: fetch p53 structure and assess quality
+pred = get_alphafold_prediction("P04637")
+print(f"Gene: {pred['gene']} ({pred['uniprotDescription']})")
+print(f"Organism: {pred['organismScientificName']}")
+print(f"Average pLDDT: {pred['globalMetricValue']}")
+print(f"Very high confidence: {pred['fractionPlddtVeryHigh']:.1%}")
+
+# Download per-residue scores
+scores = get_confidence_scores(pred)
+high_conf = [i+1 for i, c in enumerate(scores["confidenceCategory"])
+             if c in ("A", "B")]
+print(f"High-confidence residues: {len(high_conf)}/{len(scores['residueNumber'])}")
+
+# Download PDB file
+path = download_structure(pred, fmt="pdb")
+print(f"Structure saved to: {path}")
+```
+
+## Academic Use Cases
+
+- **Drug target assessment**: Check pLDDT scores in binding pockets before docking
+- **Homology model comparison**: Compare AlphaFold predictions with experimental PDB structures
+- **Disorder prediction**: Low pLDDT regions (<50) correlate with intrinsically disordered regions
+- **Variant interpretation**: Use AlphaMissense annotations (via `amAnnotationsUrl`) to assess pathogenicity
+- **Structural coverage**: Quickly check if a protein of interest has a predicted structure
+- **Batch proteome analysis**: Retrieve predictions for all proteins in a reference proteome
+
+## References
+
+- [AlphaFold Protein Structure Database](https://alphafold.ebi.ac.uk/)
+- [AlphaFold DB API Documentation](https://alphafold.ebi.ac.uk/api/openapi.json)
+- [AlphaFold DB FTP (Bulk Downloads)](https://ftp.ebi.ac.uk/pub/databases/alphafold/)
+- Jumper, J. et al. (2021). "Highly accurate protein structure prediction with AlphaFold." *Nature* 596, 583-589.
+- Varadi, M. et al. (2022). "AlphaFold Protein Structure Database: massively expanding the structural coverage of protein-sequence space with high-accuracy models." *Nucleic Acids Research* 50(D1).