@nahisaho/satori 0.17.0 → 0.19.0

package/src/.github/skills/scientific-crossref-metadata/SKILL.md

@@ -0,0 +1,313 @@
+---
+name: scientific-crossref-metadata
+description: |
+  CrossRef メタデータスキル。CrossRef REST API による
+  DOI 解決・論文メタデータ・引用数・ジャーナル情報・
+  助成金情報検索。ToolUniverse 連携: crossref。
+tu_tools:
+  - key: crossref
+    name: CrossRef
+    description: DOI 解決・論文メタデータ・引用数・ジャーナル情報
+---
+
+# Scientific CrossRef Metadata
+
+CrossRef REST API を活用した学術文献 DOI 解決・メタデータ検索・
+引用分析・ジャーナル情報・助成金レジストリ検索パイプラインを
+提供する。
+
+## When to Use
+
+- DOI から論文メタデータを取得するとき
+- 学術文献をタイトル・著者で検索するとき
+- ジャーナルの ISSN やインパクト情報を調べるとき
+- 論文の引用数・被引用数を確認するとき
+- 研究助成金の情報を検索するとき
+- 参考文献リストのメタデータを一括取得するとき
+- 特定出版社やジャーナルの出版傾向を分析するとき
+
+---
+
+## Quick Start
+
+## 1. DOI 解決・論文メタデータ
+
+```python
+import requests
+import pandas as pd
+
+CR_BASE = "https://api.crossref.org"
+CR_HEADERS = {
+    "User-Agent": "SATORI/0.18.0 (mailto:your@email.com)",
+}
+
+
+def crossref_resolve_doi(doi):
+    """
+    CrossRef — DOI からメタデータ取得。
+
+    Parameters:
+        doi: str — DOI (例: "10.1038/s41586-020-2649-2")
+    """
+    url = f"{CR_BASE}/works/{doi}"
+    resp = requests.get(url, headers=CR_HEADERS, timeout=30)
+    resp.raise_for_status()
+    item = resp.json().get("message", {})
+
+    authors = []
+    for a in item.get("author", []):
+        name = f"{a.get('given', '')} {a.get('family', '')}"
+        authors.append(name.strip())
+
+    result = {
+        "doi": item.get("DOI", ""),
+        "title": " ".join(item.get("title", [])),
+        "authors": "; ".join(authors[:10]),
+        "journal": " ".join(
+            item.get("container-title", [])),
+        "publisher": item.get("publisher", ""),
+        "type": item.get("type", ""),
+        "published_date": _cr_date(
+            item.get("published-print") or
+            item.get("published-online")),
+        "citation_count": item.get(
+            "is-referenced-by-count", 0),
+        "reference_count": item.get("reference-count", 0),
+        "issn": ", ".join(item.get("ISSN", [])),
+        "url": item.get("URL", ""),
+        "abstract": (item.get("abstract") or "")[:500],
+        "funder": "; ".join(
+            f.get("name", "")
+            for f in item.get("funder", [])),
+        "license": _cr_license(item),
+    }
+
+    print(f"CrossRef DOI: {doi}")
+    print(f"  {result['title'][:80]}")
+    print(f"  Citations: {result['citation_count']}")
+    return result
+
+
+def _cr_date(date_obj):
+    if not date_obj:
+        return ""
+    parts = date_obj.get("date-parts", [[]])[0]
+    return "-".join(str(p) for p in parts)
+
+
+def _cr_license(item):
+    licenses = item.get("license", [])
+    if licenses:
+        return licenses[0].get("content-version", "")
+    return ""
+```
+
+## 2. 論文検索
+
+```python
+def crossref_search_works(query, limit=50,
+                             sort="relevance",
+                             filter_type=None,
+                             from_date=None):
+    """
+    CrossRef — 論文検索。
+
+    Parameters:
+        query: str — 検索クエリ
+        limit: int — 最大結果数
+        sort: str — ソート ("relevance", "published",
+            "is-referenced-by-count")
+        filter_type: str — 文献タイプフィルタ
+            (例: "journal-article")
+        from_date: str — 開始日 (例: "2020-01-01")
+    """
+    url = f"{CR_BASE}/works"
+    params = {
+        "query": query,
+        "rows": min(limit, 1000),
+        "sort": sort,
+    }
+
+    filters = []
+    if filter_type:
+        filters.append(f"type:{filter_type}")
+    if from_date:
+        filters.append(f"from-pub-date:{from_date}")
+    if filters:
+        params["filter"] = ",".join(filters)
+
+    resp = requests.get(url, params=params,
+                        headers=CR_HEADERS, timeout=30)
+    resp.raise_for_status()
+    data = resp.json().get("message", {})
+
+    results = []
+    for item in data.get("items", []):
+        authors = []
+        for a in item.get("author", []):
+            name = f"{a.get('given', '')} {a.get('family', '')}"
+            authors.append(name.strip())
+        results.append({
+            "doi": item.get("DOI", ""),
+            "title": " ".join(item.get("title", [])),
+            "authors": "; ".join(authors[:5]),
+            "journal": " ".join(
+                item.get("container-title", [])),
+            "year": _cr_date(
+                item.get("published-print") or
+                item.get("published-online")),
+            "citations": item.get(
+                "is-referenced-by-count", 0),
+            "type": item.get("type", ""),
+        })
+
+    df = pd.DataFrame(results)
+    total = data.get("total-results", 0)
+    print(f"CrossRef search: {len(df)}/{total} works "
+          f"(query='{query}')")
+    return df
+```
+
+## 3. ジャーナル情報・助成金検索
+
+```python
+def crossref_journal_info(issn):
+    """
+    CrossRef — ジャーナル情報取得。
+
+    Parameters:
+        issn: str — ISSN (例: "0028-0836")
+    """
+    url = f"{CR_BASE}/journals/{issn}"
+    resp = requests.get(url, headers=CR_HEADERS, timeout=30)
+    resp.raise_for_status()
+    data = resp.json().get("message", {})
+
+    counts = data.get("counts", {})
+    result = {
+        "issn": issn,
+        "title": data.get("title", ""),
+        "publisher": data.get("publisher", ""),
+        "subjects": "; ".join(
+            s.get("name", "")
+            for s in data.get("subjects", [])),
+        "total_dois": counts.get("total-dois", 0),
+        "current_dois": counts.get("current-dois", 0),
+        "backfile_dois": counts.get("backfile-dois", 0),
+    }
+
+    print(f"CrossRef journal: {result['title']} "
+          f"({result['total_dois']} DOIs)")
+    return result
+
+
+def crossref_search_funders(query, limit=20):
+    """
+    CrossRef — 助成金機関検索。
+
+    Parameters:
+        query: str — 機関名 (例: "NIH", "JSPS")
+        limit: int — 最大結果数
+    """
+    url = f"{CR_BASE}/funders"
+    params = {"query": query, "rows": limit}
+    resp = requests.get(url, params=params,
+                        headers=CR_HEADERS, timeout=30)
+    resp.raise_for_status()
+    data = resp.json().get("message", {})
+
+    results = []
+    for item in data.get("items", []):
+        results.append({
+            "funder_id": item.get("id", ""),
+            "name": item.get("name", ""),
+            "location": item.get("location", ""),
+            "alt_names": "; ".join(
+                item.get("alt-names", [])[:3]),
+            "work_count": item.get("work-count", 0),
+        })
+
+    df = pd.DataFrame(results)
+    print(f"CrossRef funders: {len(df)} (query='{query}')")
+    return df
+```
+
+## 4. CrossRef 統合パイプライン
+
+```python
+def crossref_pipeline(query, dois=None,
+                         output_dir="results"):
+    """
+    CrossRef 統合パイプライン。
+
+    Parameters:
+        query: str — 検索クエリ
+        dois: list[str] — DOI リスト (直接解決)
+        output_dir: str — 出力ディレクトリ
+    """
+    from pathlib import Path
+    output_dir = Path(output_dir)
+    output_dir.mkdir(parents=True, exist_ok=True)
+
+    # 1) 論文検索
+    works = crossref_search_works(query)
+    works.to_csv(output_dir / "works.csv", index=False)
+
+    # 2) DOI 一括解決
+    if dois:
+        resolved = []
+        for doi in dois:
+            try:
+                meta = crossref_resolve_doi(doi)
+                resolved.append(meta)
+            except Exception as e:
+                print(f"  Warning: {doi} — {e}")
+                continue
+        resolved_df = pd.DataFrame(resolved)
+        resolved_df.to_csv(output_dir / "doi_resolved.csv",
+                           index=False)
+
+    # 3) 引用分析
+    if not works.empty:
+        stats = {
+            "total_works": len(works),
+            "total_citations": works["citations"].sum(),
+            "mean_citations": works["citations"].mean(),
+            "median_citations": works["citations"].median(),
+            "max_citations": works["citations"].max(),
+        }
+        pd.DataFrame([stats]).to_csv(
+            output_dir / "citation_stats.csv", index=False)
+
+    print(f"CrossRef pipeline: {output_dir}")
+    return {"works": works}
+```
+
+---
+
+## ToolUniverse 連携
+
+| TU Key | ツール名 | 連携内容 |
+|--------|---------|---------|
+| `crossref` | CrossRef | DOI 解決・メタデータ・引用・ジャーナル情報 |
+
+## パイプライン統合
+
+```
+literature-search → crossref-metadata → citation-checker
+  (PubMed/NCBI)   (CrossRef REST API)  (引用品質検証)
+       │                  │                  ↓
+  semantic-scholar ──────┘          deep-research
+  (S2 Academic Graph)   │          (知識統合)
+                        ↓
+               bibliometrics
+               (書誌計量分析)
+```
+
+## パイプライン出力
+
+| ファイル | 説明 | 次スキル |
+|---------|------|---------|
+| `results/works.csv` | 論文検索結果 | → semantic-scholar |
+| `results/doi_resolved.csv` | DOI メタデータ | → citation-checker |
+| `results/citation_stats.csv` | 引用統計 | → bibliometrics |

package/src/.github/skills/scientific-depmap-dependencies/SKILL.md

@@ -0,0 +1,239 @@
+---
+name: scientific-depmap-dependencies
+description: |
+  DepMap 依存性スキル。Cancer Dependency Map (DepMap) Portal
+  API によるがん細胞株 CRISPR/RNAi 依存性スコア・薬剤
+  感受性データ・遺伝子効果取得。ToolUniverse 連携: depmap。
+tu_tools:
+  - key: depmap
+    name: DepMap
+    description: がん細胞株依存性マップ API
+---
+
+# Scientific DepMap Dependencies
+
+Cancer Dependency Map (DepMap) Portal API を活用した
+がん細胞株の CRISPR/RNAi 遺伝子依存性スコア取得・
+薬剤感受性データ・遺伝子効果パイプラインを提供する。
+
+## When to Use
+
+- がん細胞株の遺伝子依存性 (CRISPR/RNAi) を調べるとき
+- 遺伝子のがん選択的必須性を評価するとき
+- 薬剤感受性データと遺伝子発現の相関を調べるとき
+- 細胞株メタデータ (組織型・疾患サブタイプ) を取得するとき
+- Cell Model Passports データを参照するとき
+
+---
+
+## Quick Start
+
+## 1. DepMap 細胞株検索・メタデータ
+
+```python
+import requests
+import pandas as pd
+
+DEPMAP_API = "https://api.cellmodelpassports.sanger.ac.uk/v1"
+
+
+def depmap_cell_lines(tissue=None, cancer_type=None,
+                         limit=50):
+    """
+    DepMap / Cell Model Passports — 細胞株検索。
+
+    Parameters:
+        tissue: str — 組織型フィルタ (例: "Breast")
+        cancer_type: str — がん種フィルタ
+            (例: "Breast Carcinoma")
+        limit: int — 最大結果数
+    """
+    url = f"{DEPMAP_API}/models"
+    params = {"page_size": limit}
+    if tissue:
+        params["tissue"] = tissue
+    if cancer_type:
+        params["cancer_type"] = cancer_type
+
+    resp = requests.get(url, params=params, timeout=30)
+    resp.raise_for_status()
+    data = resp.json()
+
+    rows = []
+    for model in data.get("data", data
+                          if isinstance(data, list)
+                          else []):
+        if isinstance(model, dict):
+            rows.append({
+                "model_id": model.get("model_id", ""),
+                "model_name": model.get("model_name", ""),
+                "tissue": model.get("tissue", ""),
+                "cancer_type": model.get(
+                    "cancer_type", ""),
+                "sample_site": model.get(
+                    "sample_site", ""),
+                "gender": model.get("gender", ""),
+            })
+
+    df = pd.DataFrame(rows[:limit])
+    print(f"DepMap cell lines: {len(df)} models")
+    return df
+```
+
+## 2. CRISPR 遺伝子依存性スコア
+
+```python
+def depmap_gene_dependency(gene_symbol,
+                              dataset="Chronos_Combined"):
+    """
+    DepMap — 遺伝子依存性スコア取得。
+
+    Parameters:
+        gene_symbol: str — 遺伝子シンボル (例: "KRAS")
+        dataset: str — データセット名
+            (例: "Chronos_Combined", "RNAi_merged")
+    """
+    # DepMap Public Portal download URL
+    DEPMAP_PORTAL = "https://depmap.org/portal/api"
+
+    url = f"{DEPMAP_PORTAL}/dataset/search"
+    params = {"gene": gene_symbol,
+              "dataset": dataset}
+
+    try:
+        resp = requests.get(url, params=params,
+                            timeout=30)
+        resp.raise_for_status()
+        data = resp.json()
+    except Exception:
+        # Fallback: alternative DepMap API
+        print(f"  DepMap portal fallback for "
+              f"{gene_symbol}")
+        data = []
+
+    rows = []
+    if isinstance(data, list):
+        for entry in data:
+            if isinstance(entry, dict):
+                rows.append({
+                    "gene": gene_symbol,
+                    "cell_line": entry.get(
+                        "cell_line_name", ""),
+                    "depmap_id": entry.get(
+                        "depmap_id", ""),
+                    "dependency_score": entry.get(
+                        "dependency", 0),
+                    "dataset": dataset,
+                })
+
+    df = pd.DataFrame(rows)
+    if not df.empty:
+        df = df.sort_values("dependency_score")
+    print(f"DepMap dependency: {gene_symbol} "
+          f"→ {len(df)} cell lines")
+    return df
+```
+
+## 3. 薬剤感受性データ
+
+```python
+def depmap_drug_sensitivity(compound_name=None,
+                               limit=100):
+    """
+    DepMap — 薬剤感受性データ取得。
+
+    Parameters:
+        compound_name: str — 化合物名フィルタ
+            (例: "Paclitaxel")
+        limit: int — 最大結果数
+    """
+    url = f"{DEPMAP_API}/drugs"
+    params = {"page_size": limit}
+    if compound_name:
+        params["name"] = compound_name
+
+    resp = requests.get(url, params=params, timeout=30)
+    resp.raise_for_status()
+    data = resp.json()
+
+    rows = []
+    for drug in data.get("data", data
+                         if isinstance(data, list)
+                         else []):
+        if isinstance(drug, dict):
+            rows.append({
+                "drug_id": drug.get("drug_id", ""),
+                "drug_name": drug.get("drug_name", ""),
+                "target": drug.get("target", ""),
+                "pathway": drug.get("pathway", ""),
+                "n_cell_lines": drug.get(
+                    "n_cell_lines_tested", 0),
+            })
+
+    df = pd.DataFrame(rows[:limit])
+    print(f"DepMap drugs: {len(df)} compounds")
+    return df
+```
+
+## 4. DepMap 統合パイプライン
+
+```python
+def depmap_pipeline(gene_symbol,
+                       output_dir="results"):
+    """
+    DepMap 統合パイプライン。
+
+    Parameters:
+        gene_symbol: str — 遺伝子シンボル
+        output_dir: str — 出力ディレクトリ
+    """
+    from pathlib import Path
+    output_dir = Path(output_dir)
+    output_dir.mkdir(parents=True, exist_ok=True)
+
+    # 1) 遺伝子依存性
+    deps = depmap_gene_dependency(gene_symbol)
+    deps.to_csv(output_dir / "depmap_dependency.csv",
+                index=False)
+
+    # 2) 関連薬剤
+    drugs = depmap_drug_sensitivity()
+    drugs.to_csv(output_dir / "depmap_drugs.csv",
+                 index=False)
+
+    # 3) 依存性の高い細胞株
+    if not deps.empty:
+        top_dependent = deps.head(10)
+        top_dependent.to_csv(
+            output_dir / "depmap_top_dependent.csv",
+            index=False)
+
+    print(f"DepMap pipeline: {gene_symbol} → {output_dir}")
+    return {"dependency": deps, "drugs": drugs}
+```
+
+---
+
+## ToolUniverse 連携
+
+| TU Key | ツール名 | 連携内容 |
+|--------|---------|---------|
+| `depmap` | DepMap | がん細胞株依存性マップ API |
+
+## パイプライン統合
+
+```
+cancer-genomics → depmap-dependencies → precision-oncology
+  (がんゲノム)       (DepMap Portal)     (精密腫瘍学)
+        │                  │                  ↓
+expression-analysis ──────┘       drug-target-profiling
+  (発現リスト)                    (標的プロファイリング)
+```
+
+## パイプライン出力
+
+| ファイル | 説明 | 次スキル |
+|---------|------|---------|
+| `results/depmap_dependency.csv` | 依存性スコア | → cancer-genomics |
+| `results/depmap_drugs.csv` | 薬剤感受性 | → drug-target-profiling |
+| `results/depmap_top_dependent.csv` | 依存細胞株 | → precision-oncology |

package/src/.github/skills/scientific-disease-research/SKILL.md

@@ -5,6 +5,10 @@ description: |
   疾患-遺伝子関連解析・希少疾患診断支援・表現型-遺伝型マッピング・
   疫学的特性評価を支援。
   「疾患と遺伝子の関連を調べて」「希少疾患を診断して」「GWAS 結果を解析して」で発火。
+tu_tools:
+  - key: disgenet
+    name: DisGeNET
+    description: 疾患-遺伝子関連スコア (GDA) データベース
 ---
 
 # Scientific Disease Research