@nahisaho/satori 0.25.4 → 0.25.5

package/bin/satori.js

@@ -4,6 +4,7 @@ const fs = require('fs');
 const path = require('path');
 
 const COMMAND = process.argv[2];
+const SUBCOMMAND = process.argv[3];
 const FLAGS = process.argv.slice(3);
 
 const PACKAGE_ROOT = path.resolve(__dirname, '..');
@@ -70,6 +71,8 @@ SATORI — Agent Skills for Science
 
 Usage:
   satori init [--force] [--dry-run]   Install .github/ skills into current directory
+  satori pipeline suggest             Interactive pipeline recommendation
+  satori pipeline list                List all available pipelines
   satori help                         Show this help message
   satori --version, -v                Show version number
 
@@ -79,6 +82,103 @@ Options:
 `);
 }
 
+// ── Pipeline Suggest ──
+
+const PIPELINES = [
+  { id: 1,  name: '仮説検証→論文化', domain: 'general',    keywords: ['仮説', '統計', '論文', 'hypothesis'], skills: 'hypothesis-engine → data-preprocessing → statistical-testing → ml-classification → publication-figures → academic-writing → critical-review' },
+  { id: 2,  name: 'バリアント→臨床', domain: 'genomics',   keywords: ['バリアント', 'variant', 'VCF', 'WGS', 'WES'], skills: 'variant-interpretation → pharmacogenomics → precision-oncology → clinical-decision-support → clinical-reporting' },
+  { id: 3,  name: 'トランスクリプトーム', domain: 'genomics', keywords: ['RNA-seq', 'トランスクリプトーム', 'DEG', '発現'], skills: 'rnaseq-analysis → pathway-enrichment → network-analysis → publication-figures' },
+  { id: 4,  name: 'エピジェネティクス', domain: 'genomics', keywords: ['エピゲノム', 'ChIP-seq', 'ATAC-seq', 'メチル化'], skills: 'epigenomics-chromatin → regulatory-genomics → noncoding-rna → gene-regulation' },
+  { id: 5,  name: 'AlphaFold 構造解析', domain: 'structural', keywords: ['AlphaFold', 'タンパク質構造', '3D', 'protein structure'], skills: 'alphafold-structures → protein-structure-analysis → molecular-docking' },
+  { id: 6,  name: 'エビデンス合成', domain: 'literature', keywords: ['メタアナリシス', 'systematic review', '文献', 'エビデンス'], skills: 'deep-research → literature-search → meta-analysis → evidence-synthesis → academic-writing → critical-review' },
+  { id: 7,  name: '創薬パイプライン', domain: 'pharma',    keywords: ['創薬', 'drug discovery', 'ADMET', 'ドッキング'], skills: 'drug-target-profiling → compound-screening → molecular-docking → admet-pharmacokinetics → drug-repurposing' },
+  { id: 8,  name: 'ML/XAI パイプライン', domain: 'ml',     keywords: ['機械学習', 'ML', 'SHAP', 'XAI', '予測モデル'], skills: 'data-preprocessing → ml-classification → ml-regression → explainable-ai → fairness-bias → publication-figures' },
+  { id: 9,  name: '環境・生態学', domain: 'ecology',       keywords: ['生態', '生物多様性', 'SDM', '環境', 'ecology'], skills: 'environmental-ecology → biodiversity-conservation → species-distribution → time-series-forecasting' },
+  { id: 10, name: '計算材料科学', domain: 'materials',     keywords: ['材料', 'materials', 'DFT', '物性'], skills: 'computational-materials → cheminformatics → molecular-dynamics → ml-regression' },
+  { id: 11, name: '医薬品安全性', domain: 'pharma',        keywords: ['有害事象', 'ファーマコビジランス', '安全性', 'adverse'], skills: 'pharmacovigilance → pharmacogenomics → regulatory-science' },
+  { id: 12, name: '希少疾患', domain: 'clinical',          keywords: ['希少疾患', 'rare disease', 'Orphanet'], skills: 'rare-disease-genetics → gene-panel-design → variant-interpretation → clinical-reporting' },
+  { id: 13, name: 'がんゲノミクス', domain: 'oncology',    keywords: ['がん', 'cancer', 'TMB', '体細胞変異'], skills: 'cancer-genomics → precision-oncology → biomarker-discovery → clinical-reporting' },
+  { id: 14, name: 'GWAS・集団遺伝学', domain: 'genomics', keywords: ['GWAS', '集団遺伝学', 'population genetics', 'biobank'], skills: 'biobank-cohort → population-genetics → statistical-testing → publication-figures' },
+  { id: 15, name: 'シングルセル', domain: 'genomics',      keywords: ['シングルセル', 'single-cell', 'scRNA-seq', '空間トランスクリプトーム'], skills: 'cellxgene-census → scvi-integration → spatial-transcriptomics → gene-regulation' },
+  { id: 16, name: 'プロテオミクス', domain: 'omics',       keywords: ['プロテオミクス', 'proteomics', '質量分析'], skills: 'proteomics → protein-structure-analysis → network-analysis' },
+  { id: 17, name: 'メタボロミクス', domain: 'omics',       keywords: ['メタボロミクス', 'metabolomics', '代謝物', '脂質'], skills: 'metabolomics → lipidomics → systems-biology → network-analysis' },
+  { id: 18, name: 'マイクロバイオーム', domain: 'ecology', keywords: ['マイクロバイオーム', 'metagenome', '16S', '腸内細菌'], skills: 'microbiome-metagenomics → metagenome-assembled-genomes → phylogenetics → environmental-ecology' },
+  { id: 19, name: 'パスウェイ・KG', domain: 'systems',     keywords: ['パスウェイ', 'ナレッジグラフ', 'knowledge graph', 'pathway'], skills: 'gene-id-mapping → pathway-enrichment → ontology-integration → network-analysis → knowledge-graph' },
+  { id: 20, name: '農業・食品', domain: 'agriculture',     keywords: ['農業', '食品', 'agriculture', 'food safety'], skills: 'agricultural-science → food-science-nutrition → environmental-ecology' },
+  { id: 21, name: '臨床情報学', domain: 'clinical',        keywords: ['臨床', 'EHR', 'FHIR', 'OMOP', '電子カルテ'], skills: 'clinical-standards → clinical-nlp → clinical-reporting → healthcare-ai → survival-clinical' },
+  { id: 22, name: 'ロボティクス・IoT', domain: 'engineering', keywords: ['ロボティクス', 'IoT', 'ロボット', 'robotics'], skills: 'robotics-automation → lab-automation → lab-data-management → interactive-dashboard' },
+  { id: 23, name: '実験計画・統計', domain: 'general',     keywords: ['実験計画', 'DOE', '検出力', 'サンプルサイズ'], skills: 'experimental-design → statistical-testing → reproducibility-assessment → publication-figures' },
+  { id: 24, name: '科学的可視化', domain: 'general',       keywords: ['可視化', 'visualization', 'ダッシュボード', 'dashboard'], skills: 'publication-figures → interactive-dashboard' },
+  { id: 25, name: '学術出版', domain: 'literature',        keywords: ['論文投稿', 'journal', 'グラント', 'grant'], skills: 'academic-writing → critical-review → citation-network' },
+  { id: 26, name: '科学教育', domain: 'education',         keywords: ['教育', 'education', 'カリキュラム'], skills: 'science-education → reproducibility-assessment' },
+];
+
+function pipelineSuggest() {
+  const readline = require('readline');
+  const rl = readline.createInterface({ input: process.stdin, output: process.stdout });
+
+  const ask = (q) => new Promise((resolve) => rl.question(q, resolve));
+
+  (async () => {
+    console.log('\n🔬 SATORI Pipeline Suggest — インタラクティブパイプライン推薦\n');
+    console.log('研究内容を入力すると、最適なパイプラインを提案します。');
+    console.log('(Ctrl+C で終了)\n');
+
+    const input = await ask('何を解析しますか？ キーワードや研究テーマを入力してください:\n> ');
+    const query = input.toLowerCase();
+
+    // Score each pipeline by keyword match
+    const scored = PIPELINES.map((p) => {
+      let score = 0;
+      for (const kw of p.keywords) {
+        if (query.includes(kw.toLowerCase())) score += 2;
+      }
+      // Partial match on name
+      if (query.includes(p.name.toLowerCase()) || p.name.toLowerCase().includes(query)) score += 1;
+      return { ...p, score };
+    })
+      .filter((p) => p.score > 0)
+      .sort((a, b) => b.score - a.score);
+
+    console.log('');
+    if (scored.length === 0) {
+      console.log('❌ 該当するパイプラインが見つかりませんでした。');
+      console.log('');
+      console.log('利用可能なキーワード例:');
+      console.log('  遺伝子/バリアント, 創薬/ADMET, RNA-seq, がん, 機械学習/ML,');
+      console.log('  メタボロミクス, マイクロバイオーム, 環境/生態, 材料, 臨床/EHR,');
+      console.log('  文献/メタアナリシス, 可視化, 論文, AlphaFold, シングルセル');
+      console.log('');
+      console.log('全パイプライン一覧は `satori pipeline list` で確認できます。');
+    } else {
+      console.log(`✅ ${scored.length} 件のパイプラインが見つかりました:\n`);
+      const top = scored.slice(0, 5);
+      for (const p of top) {
+        console.log(`  📋 Pipeline #${p.id}: ${p.name}`);
+        console.log(`     スキル連鎖: ${p.skills}`);
+        console.log('');
+      }
+      if (scored.length > 5) {
+        console.log(`  ... 他 ${scored.length - 5} 件`);
+      }
+      console.log('詳細は docs/SATORI_PIPELINE_EXAMPLES.md を参照してください。');
+    }
+
+    rl.close();
+  })();
+}
+
+function pipelineList() {
+  console.log('\n📋 SATORI パイプライン一覧 (26 ドメインパイプライン)\n');
+  for (const p of PIPELINES) {
+    console.log(`  #${String(p.id).padStart(2, ' ')}  ${p.name}`);
+    console.log(`       ${p.skills}`);
+    console.log('');
+  }
+  console.log('クロスドメイン (15), 産業特化 (5), 方法論特化 (4) パイプラインは');
+  console.log('docs/SATORI_PIPELINE_EXAMPLES.md を参照してください。');
+}
+
 function showVersion() {
   const pkg = require(path.join(PACKAGE_ROOT, 'package.json'));
   console.log(pkg.version);
@@ -88,6 +188,17 @@ switch (COMMAND) {
   case 'init':
     init();
     break;
+  case 'pipeline':
+    if (SUBCOMMAND === 'suggest') {
+      pipelineSuggest();
+    } else if (SUBCOMMAND === 'list') {
+      pipelineList();
+    } else {
+      console.error(`Unknown pipeline subcommand: ${SUBCOMMAND || '(none)'}`);
+      console.log('Usage: satori pipeline suggest | satori pipeline list');
+      process.exit(1);
+    }
+    break;
   case 'help':
   case '--help':
   case '-h':

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@nahisaho/satori",
-  "version": "0.25.4",
+  "version": "0.25.5",
   "description": "SATORI — Agent Skills for Science. GitHub Copilot Agent Skills collection for scientific data analysis.",
   "main": "index.js",
   "bin": {

package/src/.github/skills/scientific-data-preprocessing/SKILL.md

@@ -399,6 +399,28 @@ def preprocessing_pipeline(df, target_col=None, config=None):
 
     print(f"  Preprocessing complete: {df.shape}")
 
+    # チェックポイント: 前処理済みデータを永続化（パイプライン連携用）
+    from pathlib import Path
+    results_dir = Path("results")
+    results_dir.mkdir(parents=True, exist_ok=True)
+    df.to_csv(results_dir / "preprocessed_data.csv", index=False)
+    print(f"  ✔ Preprocessed data saved: results/preprocessed_data.csv ({df.shape})")
+
+    # 前処理サマリーをJSONに保存
+    import json
+    summary = {
+        "original_shape": [n_original, len(df.columns)],
+        "processed_shape": list(df.shape),
+        "duplicates_removed": n_original - len(df) if config.get("drop_duplicates", True) else 0,
+        "numeric_columns": len(df.select_dtypes(include=[np.number]).columns),
+        "categorical_columns": len(df.select_dtypes(include=["object", "category"]).columns),
+        "scaling_method": config.get("scaling_method", "standard"),
+        "missing_strategy": config.get("missing_strategy", "auto"),
+    }
+    with open(results_dir / "preprocessing_summary.json", "w") as f:
+        json.dump(summary, f, indent=2, ensure_ascii=False)
+    print(f"  ✔ Preprocessing summary saved: results/preprocessing_summary.json")
+
     return df, {"encoders": encoders, "scaler": scaler}
 ```
 

package/src/.github/skills/scientific-metagenome-assembled-genomes/SKILL.md

@@ -268,6 +268,46 @@ def mag_pipeline(assembly_fasta, bam_file,
           f"{len(quality)} QC passed → "
           f"{len(derep)} dereplicated")
 
+    # === パイプライン連携用の構造化出力 ===
+    results = Path(output_dir) / "results"
+    results.mkdir(parents=True, exist_ok=True)
+
+    # 1) MAG品質サマリーCSV (→ phylogenetics, environmental-ecology)
+    quality.to_csv(results / "mag_quality_summary.csv", index=False)
+    print(f"  ✔ MAG quality summary: {results / 'mag_quality_summary.csv'}")
+
+    # 2) 分類学サマリーCSV (→ phylogenetics)
+    if not taxonomy.empty:
+        taxonomy.to_csv(results / "mag_taxonomy.csv", index=False)
+        print(f"  ✔ MAG taxonomy: {results / 'mag_taxonomy.csv'}")
+
+    # 3) 代表MAGをFASTAに統合 (→ phylogenetics, annotation)
+    representative_fasta = results / "representative_mags.fasta"
+    with open(representative_fasta, "w") as f:
+        for mag_path in derep:
+            mag_name = Path(mag_path).stem
+            with open(mag_path) as mag_f:
+                for line in mag_f:
+                    if line.startswith(">"):
+                        f.write(f">{mag_name}_{line[1:]}")
+                    else:
+                        f.write(line)
+    print(f"  ✔ Representative MAGs FASTA: {representative_fasta}")
+
+    # 4) パイプラインサマリーJSON
+    import json
+    pipeline_summary = {
+        "total_bins": len(bins),
+        "quality_passed": len(quality),
+        "high_quality": int((quality["quality"] == "high").sum()) if "quality" in quality.columns else 0,
+        "medium_quality": int((quality["quality"] == "medium").sum()) if "quality" in quality.columns else 0,
+        "dereplicated": len(derep),
+        "classified": len(taxonomy) if not taxonomy.empty else 0,
+    }
+    with open(results / "mag_pipeline_summary.json", "w") as f:
+        json.dump(pipeline_summary, f, indent=2)
+    print(f"  ✔ Pipeline summary: {results / 'mag_pipeline_summary.json'}")
+
     return {
         "bins": bins,
         "checkm": checkm,