@nahisaho/satori 0.17.0 → 0.19.0

package/src/.github/skills/scientific-icgc-cancer-data/SKILL.md

@@ -0,0 +1,351 @@
+---
+name: scientific-icgc-cancer-data
+description: |
+  ICGC がんゲノムデータスキル。ICGC ARGO DCC API および
+  レガシー API による国際がんゲノムデータ検索・ドナー/
+  検体/変異解析。直接 API (ToolUniverse 非連携)。
+tu_tools: []
+---
+
+# Scientific ICGC Cancer Data
+
+ICGC (International Cancer Genome Consortium) ARGO DCC API を
+活用した国際がんゲノムデータ検索・変異統計・がん種横断解析
+パイプラインを提供する。
+
+## When to Use
+
+- 国際がんゲノムプロジェクトのデータを検索するとき
+- がん種ごとの体細胞変異プロファイルを調べるとき
+- ドナー・検体・変異の統計情報を取得するとき
+- がんゲノムの変異シグネチャを分析するとき
+- PCAWG (Pan-Cancer Analysis of Whole Genomes) データを活用するとき
+- がん遺伝子変異の国際比較データが必要なとき
+
+---
+
+## Quick Start
+
+## 1. ICGC プロジェクト・ドナー検索
+
+```python
+import requests
+import pandas as pd
+
+ICGC_BASE = "https://dcc.icgc.org/api/v1"
+
+
+def icgc_search_projects(query=None, limit=50):
+    """
+    ICGC — がんゲノムプロジェクト検索。
+
+    Parameters:
+        query: str — 検索キーワード (例: "lung", "BRCA")
+        limit: int — 最大結果数
+    """
+    url = f"{ICGC_BASE}/projects"
+    params = {"size": limit, "from": 1}
+    if query:
+        params["filters"] = (
+            f'{{"project":{{"primarySite":'
+            f'{{"is":["{query}"]}}}}}}'
+        )
+
+    resp = requests.get(url, params=params, timeout=30)
+    resp.raise_for_status()
+    data = resp.json()
+
+    results = []
+    for hit in data.get("hits", []):
+        results.append({
+            "project_id": hit.get("id", ""),
+            "project_name": hit.get("name", ""),
+            "primary_site": hit.get("primarySite", ""),
+            "tumour_type": hit.get("tumourType", ""),
+            "tumour_subtype": hit.get("tumourSubtype", ""),
+            "primary_country": "; ".join(
+                hit.get("primaryCountries", [])),
+            "total_donors": hit.get("totalDonorCount", 0),
+            "ssm_count": hit.get("ssmCount", 0),
+        })
+
+    df = pd.DataFrame(results)
+    if not df.empty:
+        df = df.sort_values("total_donors", ascending=False)
+
+    total = data.get("pagination", {}).get("total", 0)
+    print(f"ICGC projects: {len(df)}/{total} "
+          f"(query='{query}')")
+    return df
+
+
+def icgc_search_donors(project_id, limit=100):
+    """
+    ICGC — プロジェクト内ドナー検索。
+
+    Parameters:
+        project_id: str — プロジェクト ID (例: "BRCA-US")
+        limit: int — 最大結果数
+    """
+    url = f"{ICGC_BASE}/donors"
+    params = {
+        "size": limit,
+        "filters": (
+            f'{{"donor":{{"projectId":'
+            f'{{"is":["{project_id}"]}}}}}}'
+        ),
+    }
+
+    resp = requests.get(url, params=params, timeout=30)
+    resp.raise_for_status()
+    data = resp.json()
+
+    results = []
+    for hit in data.get("hits", []):
+        results.append({
+            "donor_id": hit.get("id", ""),
+            "project_id": project_id,
+            "primary_site": hit.get("primarySite", ""),
+            "gender": hit.get("gender", ""),
+            "vital_status": hit.get("vitalStatus", ""),
+            "age_at_diagnosis": hit.get("ageAtDiagnosis"),
+            "disease_status": hit.get(
+                "diseaseStatusLastFollowup", ""),
+            "ssm_count": hit.get("ssmCount", 0),
+        })
+
+    df = pd.DataFrame(results)
+    total = data.get("pagination", {}).get("total", 0)
+    print(f"ICGC donors: {len(df)}/{total} "
+          f"(project={project_id})")
+    return df
+```
+
+## 2. 体細胞変異 (SSM) 検索
+
+```python
+def icgc_search_mutations(gene_symbol=None,
+                             project_id=None,
+                             consequence_type=None,
+                             limit=100):
+    """
+    ICGC — 体細胞変異 (Simple Somatic Mutation) 検索。
+
+    Parameters:
+        gene_symbol: str — 遺伝子シンボル (例: "TP53")
+        project_id: str — プロジェクト ID
+        consequence_type: str — 変異タイプ
+            (例: "missense_variant")
+        limit: int — 最大結果数
+    """
+    url = f"{ICGC_BASE}/mutations"
+    filters = {}
+
+    if gene_symbol:
+        filters["gene"] = {"symbol": {"is": [gene_symbol]}}
+    if project_id:
+        filters["donor"] = {"projectId": {"is": [project_id]}}
+    if consequence_type:
+        filters["mutation"] = {
+            "consequenceType": {"is": [consequence_type]}
+        }
+
+    import json
+    params = {
+        "size": limit,
+        "filters": json.dumps(filters) if filters else "{}",
+    }
+
+    resp = requests.get(url, params=params, timeout=30)
+    resp.raise_for_status()
+    data = resp.json()
+
+    results = []
+    for hit in data.get("hits", []):
+        # 主要な consequence 取得
+        consequences = hit.get("consequences", [])
+        top_cons = consequences[0] if consequences else {}
+
+        results.append({
+            "mutation_id": hit.get("id", ""),
+            "chromosome": hit.get("chromosome", ""),
+            "start": hit.get("start"),
+            "end": hit.get("end"),
+            "mutation": hit.get("mutation", ""),
+            "type": hit.get("type", ""),
+            "gene_symbol": top_cons.get("geneSymbol", ""),
+            "consequence_type": top_cons.get("type", ""),
+            "aa_mutation": top_cons.get("aaMutation", ""),
+            "affected_donors": hit.get(
+                "affectedDonorCountTotal", 0),
+            "affected_projects": hit.get(
+                "affectedProjectCount", 0),
+            "functional_impact": hit.get(
+                "functionalImpact", ""),
+        })
+
+    df = pd.DataFrame(results)
+    if not df.empty:
+        df = df.sort_values("affected_donors",
+                            ascending=False)
+
+    total = data.get("pagination", {}).get("total", 0)
+    print(f"ICGC mutations: {len(df)}/{total} "
+          f"(gene={gene_symbol}, project={project_id})")
+    return df
+```
+
+## 3. がん種統計・変異サマリー
+
+```python
+def icgc_cancer_stats(project_id=None):
+    """
+    ICGC — がん種統計サマリー。
+
+    Parameters:
+        project_id: str — プロジェクト ID (None で全体統計)
+    """
+    if project_id:
+        url = f"{ICGC_BASE}/projects/{project_id}"
+        resp = requests.get(url, timeout=30)
+        resp.raise_for_status()
+        data = resp.json()
+
+        stats = {
+            "project_id": project_id,
+            "project_name": data.get("name", ""),
+            "primary_site": data.get("primarySite", ""),
+            "total_donors": data.get("totalDonorCount", 0),
+            "total_specimens": data.get(
+                "totalSpecimenCount", 0),
+            "ssm_count": data.get("ssmCount", 0),
+            "repository": "; ".join(
+                data.get("repository", [])),
+        }
+        print(f"ICGC stats: {project_id} — "
+              f"{stats['total_donors']} donors, "
+              f"{stats['ssm_count']} mutations")
+        return stats
+    else:
+        # 全プロジェクト概要
+        projects = icgc_search_projects(limit=200)
+        summary = {
+            "total_projects": len(projects),
+            "total_donors": projects[
+                "total_donors"].sum(),
+            "total_ssm": projects["ssm_count"].sum(),
+            "top_sites": projects.groupby(
+                "primary_site")["total_donors"].sum(
+                ).sort_values(ascending=False).head(10
+                ).to_dict(),
+        }
+        print(f"ICGC summary: {summary['total_projects']} "
+              f"projects, {summary['total_donors']} donors")
+        return summary
+
+
+def icgc_gene_mutation_frequency(gene_symbol, top_n=20):
+    """
+    ICGC — 遺伝子別がん種変異頻度。
+
+    Parameters:
+        gene_symbol: str — 遺伝子シンボル
+        top_n: int — 上位がん種数
+    """
+    mutations = icgc_search_mutations(
+        gene_symbol=gene_symbol, limit=500)
+
+    if mutations.empty:
+        return pd.DataFrame()
+
+    # プロジェクト別集計
+    freq = mutations.groupby("gene_symbol").agg(
+        total_mutations=("mutation_id", "count"),
+        total_affected_donors=("affected_donors", "sum"),
+        mutation_types=("consequence_type",
+                        lambda x: "; ".join(x.unique()[:5])),
+    ).reset_index()
+
+    print(f"ICGC gene frequency: {gene_symbol} — "
+          f"{len(freq)} entries")
+    return freq
+```
+
+## 4. ICGC 統合パイプライン
+
+```python
+def icgc_pipeline(gene_symbols, cancer_site=None,
+                     output_dir="results"):
+    """
+    ICGC 統合パイプライン。
+
+    Parameters:
+        gene_symbols: list[str] — 遺伝子リスト
+        cancer_site: str — がん部位フィルタ
+        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) プロジェクト検索
+    projects = icgc_search_projects(query=cancer_site)
+    projects.to_csv(output_dir / "projects.csv", index=False)
+
+    # 2) 遺伝子別変異検索
+    all_mutations = []
+    for gene in gene_symbols:
+        try:
+            muts = icgc_search_mutations(
+                gene_symbol=gene, limit=200)
+            muts["query_gene"] = gene
+            all_mutations.append(muts)
+        except Exception as e:
+            print(f"  Warning: {gene} — {e}")
+            continue
+
+    if all_mutations:
+        combined = pd.concat(all_mutations,
+                              ignore_index=True)
+        combined.to_csv(output_dir / "mutations.csv",
+                        index=False)
+
+    # 3) がん種統計
+    if not projects.empty:
+        top_project = projects.iloc[0]["project_id"]
+        stats = icgc_cancer_stats(project_id=top_project)
+        pd.DataFrame([stats]).to_csv(
+            output_dir / "cancer_stats.csv", index=False)
+
+    print(f"ICGC pipeline: {output_dir}")
+    return {"projects": projects}
+```
+
+---
+
+## ToolUniverse 連携
+
+| TU Key | ツール名 | 連携内容 |
+|--------|---------|---------|
+| (direct) | ICGC DCC API | 直接 REST API — TU 非連携 |
+
+## パイプライン統合
+
+```
+cancer-genomics → icgc-cancer-data → precision-oncology
+  (がんゲノム全般)  (ICGC DCC API)   (精密腫瘍学)
+       │                 │                 ↓
+  tcga-data ────────────┘        clinical-decision-support
+  (TCGA データ)      │            (臨床意思決定)
+                     ↓
+          variant-interpretation
+          (変異臨床解釈)
+```
+
+## パイプライン出力
+
+| ファイル | 説明 | 次スキル |
+|---------|------|---------|
+| `results/projects.csv` | プロジェクト一覧 | → cancer-genomics |
+| `results/mutations.csv` | 体細胞変異 | → variant-interpretation |
+| `results/cancer_stats.csv` | がん種統計 | → precision-oncology |

package/src/.github/skills/scientific-metabolomics-databases/SKILL.md

@@ -6,6 +6,10 @@ description: |
   (NIH メタボロミクスリポジトリ) の 3 大メタボロミクス DB を統合した
   代謝物同定、パスウェイマッピング、バイオマーカー発見、
   RefMet 標準化命名パイプライン。13 の ToolUniverse SMCP ツールと連携。
+tu_tools:
+  - key: metacyc
+    name: MetaCyc
+    description: 代謝パスウェイ・反応・化合物データベース
 ---
 
 # Scientific Metabolomics Databases

package/src/.github/skills/scientific-opentargets-genetics/SKILL.md

@@ -0,0 +1,299 @@
+---
+name: scientific-opentargets-genetics
+description: |
+  Open Targets Platform 遺伝学スキル。Open Targets Platform
+  GraphQL API を用いた標的-疾患アソシエーション・薬剤
+  エビデンス・L2G 遺伝的関連・ファーマコゲノミクス検索。
+  ToolUniverse 連携: opentarget。
+tu_tools:
+  - key: opentarget
+    name: Open Targets
+    description: 標的-疾患アソシエーション GraphQL API
+---
+
+# Scientific Open Targets Genetics
+
+Open Targets Platform GraphQL API を活用した標的-疾患
+アソシエーションスコア取得・薬剤エビデンス検索・L2G
+遺伝的関連パイプラインを提供する。
+
+## When to Use
+
+- 遺伝子 (標的) と疾患のアソシエーションスコアを検索するとき
+- 薬剤エビデンスデータを取得するとき
+- GWAS バリアントから遺伝子を L2G スコアでマッピングするとき
+- 標的の安全性プロファイルを確認するとき
+- ファーマコゲノミクスデータを検索するとき
+
+---
+
+## Quick Start
+
+## 1. 標的-疾患アソシエーション
+
+```python
+import requests
+import pandas as pd
+
+OT_API = ("https://api.platform.opentargets.org"
+          "/api/v4/graphql")
+
+
+def ot_target_disease_assoc(target_id, limit=25):
+    """
+    Open Targets — 標的-疾患アソシエーション。
+
+    Parameters:
+        target_id: str — Ensembl Gene ID
+            (例: "ENSG00000012048" = BRCA1)
+        limit: int — 最大結果数
+    """
+    query = """
+    query targetDisease($id: String!, $size: Int!) {
+      target(ensemblId: $id) {
+        id
+        approvedSymbol
+        associatedDiseases(page: {size: $size, index: 0}) {
+          count
+          rows {
+            disease { id name }
+            score
+            datatypeScores {
+              componentId: id
+              score
+            }
+          }
+        }
+      }
+    }
+    """
+    variables = {"id": target_id, "size": limit}
+    resp = requests.post(OT_API,
+                         json={"query": query,
+                               "variables": variables},
+                         timeout=30)
+    resp.raise_for_status()
+    data = resp.json()["data"]["target"]
+
+    rows = []
+    for r in data["associatedDiseases"]["rows"]:
+        row = {
+            "target_id": target_id,
+            "target_symbol": data["approvedSymbol"],
+            "disease_id": r["disease"]["id"],
+            "disease_name": r["disease"]["name"],
+            "overall_score": r["score"],
+        }
+        for dt in r["datatypeScores"]:
+            row[dt["componentId"]] = dt["score"]
+        rows.append(row)
+
+    df = pd.DataFrame(rows)
+    total = data["associatedDiseases"]["count"]
+    print(f"OT associations: {data['approvedSymbol']} "
+          f"→ {len(df)}/{total} diseases")
+    return df
+```
+
+## 2. 薬剤エビデンス
+
+```python
+def ot_drug_evidence(target_id, disease_id, limit=50):
+    """
+    Open Targets — 薬剤エビデンス。
+
+    Parameters:
+        target_id: str — Ensembl Gene ID
+        disease_id: str — EFO Disease ID
+            (例: "EFO_0000305" = breast carcinoma)
+        limit: int — 最大結果数
+    """
+    query = """
+    query drugEvidence($ensemblId: String!,
+                       $efoId: String!,
+                       $size: Int!) {
+      disease(efoId: $efoId) {
+        id
+        name
+        evidences(
+          ensemblIds: [$ensemblId]
+          datasourceIds: ["chembl"]
+          size: $size
+        ) {
+          count
+          rows {
+            id
+            score
+            drug {
+              id name drugType
+              maximumClinicalTrialPhase
+              mechanismsOfAction {
+                rows { actionType }
+              }
+            }
+            clinicalPhase
+            clinicalStatus
+            urls { niceName url }
+          }
+        }
+      }
+    }
+    """
+    variables = {"ensemblId": target_id,
+                 "efoId": disease_id,
+                 "size": limit}
+    resp = requests.post(OT_API,
+                         json={"query": query,
+                               "variables": variables},
+                         timeout=30)
+    resp.raise_for_status()
+    data = resp.json()["data"]["disease"]
+
+    results = []
+    for ev in data["evidences"]["rows"]:
+        drug = ev.get("drug", {})
+        moas = drug.get("mechanismsOfAction", {})
+        moa_list = [m["actionType"]
+                    for m in moas.get("rows", [])]
+        results.append({
+            "disease": data["name"],
+            "drug_id": drug.get("id", ""),
+            "drug_name": drug.get("name", ""),
+            "drug_type": drug.get("drugType", ""),
+            "max_phase": drug.get(
+                "maximumClinicalTrialPhase", 0),
+            "clinical_phase": ev.get("clinicalPhase", ""),
+            "clinical_status": ev.get(
+                "clinicalStatus", ""),
+            "moa": "; ".join(moa_list),
+            "score": ev.get("score", 0),
+        })
+
+    df = pd.DataFrame(results)
+    print(f"OT drug evidence: {len(df)} entries")
+    return df
+```
+
+## 3. L2G 遺伝的関連 (Locus-to-Gene)
+
+```python
+def ot_l2g_variants(study_id, limit=50):
+    """
+    Open Targets Genetics — L2G バリアント-遺伝子マッピング。
+
+    Parameters:
+        study_id: str — GWAS Study ID
+            (例: "GCST004988")
+        limit: int — 最大結果数
+    """
+    # OT Genetics API
+    OT_GENETICS = ("https://api.genetics.opentargets.org"
+                   "/graphql")
+    query = """
+    query l2g($studyId: String!, $size: Int!) {
+      studyLocus2GeneTable(studyId: $studyId,
+                           pageSize: $size) {
+        rows {
+          gene { id symbol }
+          variant { id rsId }
+          yProbaModel
+          yProbaDistance
+          yProbaInteraction
+          yProbaMolecularQTL
+          yProbaPathogenicity
+          hasColoc
+          distanceToLocus
+        }
+      }
+    }
+    """
+    variables = {"studyId": study_id, "size": limit}
+    resp = requests.post(OT_GENETICS,
+                         json={"query": query,
+                               "variables": variables},
+                         timeout=30)
+    resp.raise_for_status()
+    data = resp.json()["data"]["studyLocus2GeneTable"]
+
+    rows = []
+    for r in data["rows"]:
+        rows.append({
+            "gene_id": r["gene"]["id"],
+            "gene_symbol": r["gene"]["symbol"],
+            "variant_id": r["variant"]["id"],
+            "rsid": r["variant"]["rsId"],
+            "l2g_score": r["yProbaModel"],
+            "distance_score": r["yProbaDistance"],
+            "interaction_score": r["yProbaInteraction"],
+            "qtl_score": r["yProbaMolecularQTL"],
+            "pathogenicity": r["yProbaPathogenicity"],
+            "has_coloc": r["hasColoc"],
+        })
+
+    df = pd.DataFrame(rows)
+    if not df.empty:
+        df = df.sort_values("l2g_score", ascending=False)
+    print(f"OT L2G: {study_id} → {len(df)} gene mappings")
+    return df
+```
+
+## 4. Open Targets 統合パイプライン
+
+```python
+def ot_pipeline(gene_symbol, ensembl_id,
+                   output_dir="results"):
+    """
+    Open Targets 統合パイプライン。
+
+    Parameters:
+        gene_symbol: str — 遺伝子シンボル (例: "BRCA1")
+        ensembl_id: str — Ensembl Gene ID
+        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) 標的-疾患アソシエーション
+    assoc = ot_target_disease_assoc(ensembl_id)
+    assoc.to_csv(output_dir / "ot_associations.csv",
+                 index=False)
+
+    # 2) トップ疾患の薬剤エビデンス
+    if not assoc.empty:
+        top_disease = assoc.iloc[0]["disease_id"]
+        drugs = ot_drug_evidence(ensembl_id, top_disease)
+        drugs.to_csv(output_dir / "ot_drugs.csv",
+                     index=False)
+
+    print(f"OT pipeline: {gene_symbol} → {output_dir}")
+    return {"associations": assoc}
+```
+
+---
+
+## ToolUniverse 連携
+
+| TU Key | ツール名 | 連携内容 |
+|--------|---------|---------|
+| `opentarget` | Open Targets | 標的-疾患アソシエーション GraphQL (~55 tools) |
+
+## パイプライン統合
+
+```
+disease-research → opentargets-genetics → drug-target-profiling
+  (疾患遺伝子)      (OT Platform API)     (標的プロファイリング)
+        │                   │                    ↓
+variant-interpretation ────┘          pharmacogenomics
+  (ClinVar/VEP)      │              (薬理ゲノミクス)
+                      ↓
+            gnomad-variants
+            (集団頻度)
+```
+
+## パイプライン出力
+
+| ファイル | 説明 | 次スキル |
+|---------|------|---------|
+| `results/ot_associations.csv` | 標的-疾患スコア | → disease-research |
+| `results/ot_drugs.csv` | 薬剤エビデンス | → drug-target-profiling |