@yangfei_93sky/biocli 0.2.0

package/dist/cli-manifest.json

@@ -0,0 +1,1500 @@
+[
+  {
+    "site": "aggregate",
+    "name": "enrichment",
+    "description": "Combined pathway enrichment from Enrichr + STRING",
+    "database": "aggregate",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "genes",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "Comma-separated gene symbols (e.g. TP53,BRCA1,EGFR,MYC,CDK2)"
+      },
+      {
+        "name": "library",
+        "type": "str",
+        "default": "KEGG_2021_Human",
+        "required": false,
+        "help": "Enrichr library (e.g. GO_Biological_Process_2023, Reactome_2022)"
+      },
+      {
+        "name": "limit",
+        "type": "int",
+        "default": 20,
+        "required": false,
+        "help": "Max results per source (1-50)"
+      },
+      {
+        "name": "species",
+        "type": "int",
+        "default": 9606,
+        "required": false,
+        "help": "NCBI taxonomy ID for STRING (default: 9606)"
+      }
+    ],
+    "columns": [
+      "term",
+      "category",
+      "source",
+      "pValue",
+      "genes"
+    ],
+    "timeout": 60,
+    "type": "ts",
+    "modulePath": "aggregate/enrichment.js"
+  },
+  {
+    "site": "aggregate",
+    "name": "gene-dossier",
+    "description": "Complete gene intelligence report (profile + literature + clinical)",
+    "database": "aggregate",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "gene",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "Gene symbol (e.g. TP53)"
+      },
+      {
+        "name": "organism",
+        "type": "str",
+        "default": "human",
+        "required": false,
+        "help": "Organism (e.g. human, mouse)"
+      },
+      {
+        "name": "papers",
+        "type": "int",
+        "default": 5,
+        "required": false,
+        "help": "Number of recent papers to include"
+      }
+    ],
+    "columns": [
+      "symbol",
+      "name",
+      "pathways",
+      "interactions",
+      "literature",
+      "clinvar"
+    ],
+    "timeout": 90,
+    "type": "ts",
+    "modulePath": "aggregate/gene-dossier.js"
+  },
+  {
+    "site": "aggregate",
+    "name": "gene-profile",
+    "description": "Complete gene profile from NCBI + UniProt + KEGG + STRING",
+    "database": "aggregate",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "genes",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "Gene symbol(s), comma-separated (e.g. TP53 or TP53,BRCA1,EGFR)"
+      },
+      {
+        "name": "organism",
+        "type": "str",
+        "default": "human",
+        "required": false,
+        "help": "Organism (e.g. human, mouse, 9606)"
+      }
+    ],
+    "columns": [
+      "symbol",
+      "name",
+      "organism",
+      "pathways",
+      "goTerms",
+      "interactions"
+    ],
+    "timeout": 60,
+    "type": "ts",
+    "modulePath": "aggregate/gene-profile.js"
+  },
+  {
+    "site": "aggregate",
+    "name": "literature-brief",
+    "description": "Literature summary with abstracts for a research topic",
+    "database": "aggregate",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "query",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "Search query (e.g. \"TP53 immunotherapy\", \"CRISPR cancer\")"
+      },
+      {
+        "name": "limit",
+        "type": "int",
+        "default": 10,
+        "required": false,
+        "help": "Number of papers (1-50)"
+      },
+      {
+        "name": "sort",
+        "type": "str",
+        "default": "relevance",
+        "required": false,
+        "help": "Sort order",
+        "choices": [
+          "relevance",
+          "date"
+        ]
+      },
+      {
+        "name": "years",
+        "type": "int",
+        "default": 5,
+        "required": false,
+        "help": "Limit to last N years"
+      }
+    ],
+    "columns": [
+      "pmid",
+      "title",
+      "journal",
+      "year",
+      "abstract"
+    ],
+    "timeout": 60,
+    "type": "ts",
+    "modulePath": "aggregate/literature-brief.js"
+  },
+  {
+    "site": "aggregate",
+    "name": "variant-dossier",
+    "description": "Comprehensive variant interpretation (dbSNP + ClinVar + VEP)",
+    "database": "aggregate",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "variant",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "Variant ID: rsID (rs334), HGVS, or genomic coordinate"
+      }
+    ],
+    "columns": [
+      "variant",
+      "gene",
+      "consequence",
+      "clinicalSignificance",
+      "condition"
+    ],
+    "timeout": 60,
+    "type": "ts",
+    "modulePath": "aggregate/variant-dossier.js"
+  },
+  {
+    "site": "aggregate",
+    "name": "variant-interpret",
+    "description": "Variant interpretation with clinical context (dbSNP + ClinVar + VEP + UniProt)",
+    "database": "aggregate",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "variant",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "Variant ID: rsID (rs334), HGVS, or genomic coordinate"
+      }
+    ],
+    "columns": [
+      "variant",
+      "gene",
+      "consequence",
+      "clinicalSignificance",
+      "interpretation"
+    ],
+    "timeout": 90,
+    "type": "ts",
+    "modulePath": "aggregate/variant-interpret.js"
+  },
+  {
+    "site": "aggregate",
+    "name": "workflow-prepare",
+    "description": "Prepare a research-ready directory from a selected dataset",
+    "database": "aggregate",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "dataset",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "GEO accession (GSE*) or SRA accession (SRR*)"
+      },
+      {
+        "name": "gene",
+        "type": "str",
+        "required": false,
+        "help": "Focus gene symbol(s), comma-separated (e.g. TP53,BRCA1)"
+      },
+      {
+        "name": "outdir",
+        "type": "str",
+        "required": true,
+        "help": "Output directory for the prepared workspace"
+      },
+      {
+        "name": "skip-download",
+        "type": "boolean",
+        "default": false,
+        "required": false,
+        "help": "Skip data download, only fetch annotations"
+      }
+    ],
+    "columns": [
+      "step",
+      "status",
+      "detail"
+    ],
+    "timeout": 300,
+    "type": "ts",
+    "modulePath": "aggregate/workflow-prepare.js"
+  },
+  {
+    "site": "aggregate",
+    "name": "workflow-scout",
+    "description": "Scout GEO/SRA datasets for a research question",
+    "database": "aggregate",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "query",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "Research topic (e.g. \"TP53 breast cancer RNA-seq\")"
+      },
+      {
+        "name": "gene",
+        "type": "str",
+        "required": false,
+        "help": "Focus gene symbol (refines search)"
+      },
+      {
+        "name": "organism",
+        "type": "str",
+        "default": "Homo sapiens",
+        "required": false,
+        "help": "Organism filter"
+      },
+      {
+        "name": "limit",
+        "type": "int",
+        "default": 10,
+        "required": false,
+        "help": "Max candidates per source"
+      },
+      {
+        "name": "type",
+        "type": "str",
+        "default": "gse",
+        "required": false,
+        "help": "GEO entry type filter",
+        "choices": [
+          "gse",
+          "gds",
+          "all"
+        ]
+      }
+    ],
+    "columns": [
+      "rank",
+      "accession",
+      "title",
+      "type",
+      "samples",
+      "date",
+      "source"
+    ],
+    "timeout": 60,
+    "type": "ts",
+    "modulePath": "aggregate/workflow-scout.js"
+  },
+  {
+    "site": "clinvar",
+    "name": "search",
+    "description": "Search ClinVar clinical variants",
+    "database": "clinvar",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "query",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "Search query (e.g. \"BRCA1\", \"rs80357906\", \"breast cancer\")"
+      },
+      {
+        "name": "limit",
+        "type": "int",
+        "default": 10,
+        "required": false,
+        "help": "Max results (1-200)"
+      }
+    ],
+    "columns": [
+      "uid",
+      "title",
+      "gene",
+      "significance",
+      "condition",
+      "accession"
+    ],
+    "type": "ts",
+    "modulePath": "clinvar/search.js"
+  },
+  {
+    "site": "clinvar",
+    "name": "variant",
+    "description": "Get ClinVar variant details by ID",
+    "database": "clinvar",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "id",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "ClinVar variation ID or accession (e.g. 37722, VCV000037722)"
+      }
+    ],
+    "columns": [
+      "uid",
+      "title",
+      "gene",
+      "significance",
+      "condition",
+      "accession",
+      "type",
+      "assembly"
+    ],
+    "type": "ts",
+    "modulePath": "clinvar/variant.js"
+  },
+  {
+    "site": "enrichr",
+    "name": "analyze",
+    "description": "Run gene set enrichment analysis",
+    "database": "enrichr",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "genes",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "Comma-separated gene symbols (e.g. TP53,BRCA1,EGFR,MYC,CDK2)"
+      },
+      {
+        "name": "library",
+        "type": "str",
+        "default": "KEGG_2021_Human",
+        "required": false,
+        "help": "Gene set library (e.g. KEGG_2021_Human, GO_Biological_Process_2023, Reactome_2022)"
+      },
+      {
+        "name": "limit",
+        "type": "int",
+        "default": 20,
+        "required": false,
+        "help": "Max results to show (1-100)"
+      }
+    ],
+    "columns": [
+      "rank",
+      "term",
+      "adjustedPValue",
+      "combinedScore",
+      "genes"
+    ],
+    "type": "ts",
+    "modulePath": "enrichr/analyze.js"
+  },
+  {
+    "site": "ensembl",
+    "name": "lookup",
+    "description": "Look up a gene by symbol in Ensembl",
+    "database": "ensembl",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "symbol",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "Gene symbol (e.g. BRCA2, TP53)"
+      },
+      {
+        "name": "species",
+        "type": "str",
+        "default": "homo_sapiens",
+        "required": false,
+        "help": "Species name (e.g. homo_sapiens, mus_musculus)"
+      }
+    ],
+    "columns": [
+      "ensemblId",
+      "symbol",
+      "biotype",
+      "chromosome",
+      "start",
+      "end",
+      "strand",
+      "description"
+    ],
+    "type": "ts",
+    "modulePath": "ensembl/lookup.js"
+  },
+  {
+    "site": "ensembl",
+    "name": "vep",
+    "description": "Predict variant effects (VEP)",
+    "database": "ensembl",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "variant",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "Variant in HGVS (e.g. \"NM_000518.5:c.20A>T\") or rsID (e.g. rs334)"
+      },
+      {
+        "name": "species",
+        "type": "str",
+        "default": "human",
+        "required": false,
+        "help": "Species (default: human)"
+      }
+    ],
+    "columns": [
+      "input",
+      "gene",
+      "consequence",
+      "impact",
+      "biotype",
+      "aminoAcid",
+      "codons"
+    ],
+    "type": "ts",
+    "modulePath": "ensembl/vep.js"
+  },
+  {
+    "site": "ensembl",
+    "name": "xrefs",
+    "description": "Get cross-references for a gene",
+    "database": "ensembl",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "symbol",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "Gene symbol (e.g. BRCA2)"
+      },
+      {
+        "name": "species",
+        "type": "str",
+        "default": "homo_sapiens",
+        "required": false,
+        "help": "Species name"
+      }
+    ],
+    "columns": [
+      "database",
+      "primaryId",
+      "displayId",
+      "description"
+    ],
+    "type": "ts",
+    "modulePath": "ensembl/xrefs.js"
+  },
+  {
+    "site": "gene",
+    "name": "fetch",
+    "description": "Download gene sequence (nucleotide or protein) in FASTA format",
+    "database": "gene",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "id",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "NCBI Gene ID (e.g. 7157) or gene symbol with --search"
+      },
+      {
+        "name": "type",
+        "type": "str",
+        "default": "nucleotide",
+        "required": false,
+        "help": "Sequence type to download",
+        "choices": [
+          "nucleotide",
+          "protein"
+        ]
+      },
+      {
+        "name": "output",
+        "type": "str",
+        "required": false,
+        "help": "Output file path (default: stdout)"
+      }
+    ],
+    "columns": [
+      "content"
+    ],
+    "type": "ts",
+    "modulePath": "gene/fetch.js"
+  },
+  {
+    "site": "gene",
+    "name": "info",
+    "description": "Get gene details by NCBI Gene ID",
+    "database": "gene",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "id",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "NCBI Gene ID (e.g. 7157 for TP53)"
+      }
+    ],
+    "columns": [
+      "geneId",
+      "symbol",
+      "name",
+      "organism",
+      "summary",
+      "chromosome",
+      "location"
+    ],
+    "type": "ts",
+    "modulePath": "gene/info.js"
+  },
+  {
+    "site": "gene",
+    "name": "search",
+    "description": "Search NCBI Gene database",
+    "database": "gene",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "query",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "Gene symbol or keyword (e.g. TP53, BRCA1)"
+      },
+      {
+        "name": "limit",
+        "type": "int",
+        "default": 10,
+        "required": false,
+        "help": "Max results (1-200)"
+      },
+      {
+        "name": "organism",
+        "type": "str",
+        "default": "human",
+        "required": false,
+        "help": "Organism name (e.g. human, mouse, rat, zebrafish)"
+      }
+    ],
+    "columns": [
+      "geneId",
+      "symbol",
+      "name",
+      "organism"
+    ],
+    "type": "ts",
+    "modulePath": "gene/search.js"
+  },
+  {
+    "site": "geo",
+    "name": "dataset",
+    "description": "Get GEO dataset details by accession",
+    "database": "gds",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "accession",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "GEO accession (e.g. GSE12345, GDS1234)"
+      }
+    ],
+    "columns": [
+      "accession",
+      "title",
+      "organism",
+      "type",
+      "platform",
+      "samples",
+      "summary",
+      "date"
+    ],
+    "type": "ts",
+    "modulePath": "geo/dataset.js"
+  },
+  {
+    "site": "geo",
+    "name": "download",
+    "description": "Download GEO supplementary files (expression matrices, etc.)",
+    "database": "gds",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "accession",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "GEO Series accession (e.g. GSE12345)"
+      },
+      {
+        "name": "outdir",
+        "type": "str",
+        "default": ".",
+        "required": false,
+        "help": "Output directory (default: current directory)"
+      },
+      {
+        "name": "list-only",
+        "type": "boolean",
+        "default": false,
+        "required": false,
+        "help": "Only list available files, do not download"
+      },
+      {
+        "name": "dry-run",
+        "type": "boolean",
+        "default": false,
+        "required": false,
+        "help": "Same as --list-only: show files without downloading"
+      },
+      {
+        "name": "pattern",
+        "type": "str",
+        "required": false,
+        "help": "Filter files by pattern (e.g. \"counts\", \"matrix\", \"tar.gz\")"
+      }
+    ],
+    "columns": [
+      "file",
+      "size",
+      "status"
+    ],
+    "type": "ts",
+    "modulePath": "geo/download.js"
+  },
+  {
+    "site": "geo",
+    "name": "samples",
+    "description": "List samples in a GEO dataset",
+    "database": "gds",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "accession",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "GEO series accession (e.g. GSE12345)"
+      },
+      {
+        "name": "limit",
+        "type": "int",
+        "default": 20,
+        "required": false,
+        "help": "Max results (1-200)"
+      }
+    ],
+    "columns": [
+      "accession",
+      "title",
+      "organism",
+      "type"
+    ],
+    "type": "ts",
+    "modulePath": "geo/samples.js"
+  },
+  {
+    "site": "geo",
+    "name": "search",
+    "description": "Search GEO datasets",
+    "database": "gds",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "query",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "Search query (e.g. \"breast cancer RNA-seq\")"
+      },
+      {
+        "name": "limit",
+        "type": "int",
+        "default": 10,
+        "required": false,
+        "help": "Max results (1-200)"
+      },
+      {
+        "name": "type",
+        "type": "str",
+        "default": "gse",
+        "required": false,
+        "help": "Entry type filter",
+        "choices": [
+          "gse",
+          "gds",
+          "gpl",
+          "gsm"
+        ]
+      }
+    ],
+    "columns": [
+      "accession",
+      "title",
+      "organism",
+      "type",
+      "samples",
+      "date"
+    ],
+    "type": "ts",
+    "modulePath": "geo/search.js"
+  },
+  {
+    "site": "kegg",
+    "name": "convert",
+    "description": "Convert IDs between KEGG and external databases",
+    "database": "kegg",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "id",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "ID to convert (e.g. hsa:7157, ncbi-geneid:7157)"
+      },
+      {
+        "name": "to",
+        "type": "str",
+        "default": "ncbi-geneid",
+        "required": false,
+        "help": "Target database",
+        "choices": [
+          "ncbi-geneid",
+          "ncbi-proteinid",
+          "uniprot"
+        ]
+      }
+    ],
+    "columns": [
+      "source",
+      "target"
+    ],
+    "type": "ts",
+    "modulePath": "kegg/convert.js"
+  },
+  {
+    "site": "kegg",
+    "name": "disease",
+    "description": "Find diseases linked to a KEGG gene",
+    "database": "kegg",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "gene",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "KEGG gene ID (e.g. hsa:7157)"
+      }
+    ],
+    "columns": [
+      "geneId",
+      "diseaseId",
+      "diseaseName"
+    ],
+    "type": "ts",
+    "modulePath": "kegg/disease.js"
+  },
+  {
+    "site": "kegg",
+    "name": "link",
+    "description": "Find KEGG cross-references for a gene",
+    "database": "kegg",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "gene",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "KEGG gene ID (e.g. hsa:7157) or comma-separated list"
+      },
+      {
+        "name": "target",
+        "type": "str",
+        "default": "pathway",
+        "required": false,
+        "help": "Target database to link to",
+        "choices": [
+          "pathway",
+          "disease",
+          "drug",
+          "compound"
+        ]
+      }
+    ],
+    "columns": [
+      "source",
+      "target"
+    ],
+    "type": "ts",
+    "modulePath": "kegg/link.js"
+  },
+  {
+    "site": "kegg",
+    "name": "pathway",
+    "description": "Get KEGG pathway details",
+    "database": "kegg",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "id",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "KEGG pathway ID (e.g. hsa05200, hsa00600)"
+      }
+    ],
+    "columns": [
+      "id",
+      "name",
+      "description",
+      "class",
+      "genes",
+      "diseases"
+    ],
+    "type": "ts",
+    "modulePath": "kegg/pathway.js"
+  },
+  {
+    "site": "pubmed",
+    "name": "abstract",
+    "description": "Get abstract text for a PubMed article",
+    "database": "pubmed",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "pmid",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "PubMed ID"
+      }
+    ],
+    "columns": [
+      "pmid",
+      "abstract"
+    ],
+    "type": "ts",
+    "modulePath": "pubmed/abstract.js"
+  },
+  {
+    "site": "pubmed",
+    "name": "cited-by",
+    "description": "Articles that cite a PubMed article",
+    "database": "pubmed",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "pmid",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "PubMed ID"
+      },
+      {
+        "name": "limit",
+        "type": "int",
+        "default": 10,
+        "required": false,
+        "help": "Max results (1-100)"
+      }
+    ],
+    "columns": [
+      "pmid",
+      "title",
+      "authors",
+      "journal",
+      "year",
+      "doi"
+    ],
+    "type": "ts",
+    "modulePath": "pubmed/cited-by.js"
+  },
+  {
+    "site": "pubmed",
+    "name": "fetch",
+    "description": "Get PubMed article details by PMID",
+    "database": "pubmed",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "pmid",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "PubMed ID (e.g. 39088800)"
+      }
+    ],
+    "columns": [
+      "pmid",
+      "title",
+      "authors",
+      "journal",
+      "year",
+      "doi",
+      "abstract"
+    ],
+    "type": "ts",
+    "modulePath": "pubmed/fetch.js"
+  },
+  {
+    "site": "pubmed",
+    "name": "info",
+    "description": "PubMed database statistics",
+    "database": "pubmed",
+    "strategy": "public",
+    "args": [],
+    "columns": [
+      "database",
+      "description",
+      "count",
+      "last_update"
+    ],
+    "pipeline": [
+      {
+        "fetch": {
+          "url": "https://eutils.ncbi.nlm.nih.gov/entrez/eutils/einfo.fcgi",
+          "params": {
+            "db": "pubmed",
+            "retmode": "json"
+          }
+        }
+      },
+      {
+        "select": "einforesult.dbinfo"
+      },
+      {
+        "map": {
+          "database": "${{ item.dbname }}",
+          "description": "${{ item.description }}",
+          "count": "${{ item.count }}",
+          "last_update": "${{ item.lastupdate }}"
+        }
+      }
+    ],
+    "type": "yaml"
+  },
+  {
+    "site": "pubmed",
+    "name": "related",
+    "description": "Find related PubMed articles",
+    "database": "pubmed",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "pmid",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "PubMed ID"
+      },
+      {
+        "name": "limit",
+        "type": "int",
+        "default": 10,
+        "required": false,
+        "help": "Max results (1-100)"
+      }
+    ],
+    "columns": [
+      "pmid",
+      "title",
+      "authors",
+      "journal",
+      "year",
+      "doi"
+    ],
+    "type": "ts",
+    "modulePath": "pubmed/related.js"
+  },
+  {
+    "site": "pubmed",
+    "name": "search",
+    "description": "Search PubMed articles",
+    "database": "pubmed",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "query",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "Search query (e.g. \"CRISPR cancer therapy\")"
+      },
+      {
+        "name": "limit",
+        "type": "int",
+        "default": 10,
+        "required": false,
+        "help": "Max results (1-200)"
+      },
+      {
+        "name": "sort",
+        "type": "str",
+        "default": "relevance",
+        "required": false,
+        "help": "Sort order",
+        "choices": [
+          "relevance",
+          "date",
+          "author",
+          "journal"
+        ]
+      }
+    ],
+    "columns": [
+      "pmid",
+      "title",
+      "authors",
+      "journal",
+      "year",
+      "doi"
+    ],
+    "type": "ts",
+    "modulePath": "pubmed/search.js"
+  },
+  {
+    "site": "snp",
+    "name": "lookup",
+    "description": "Look up SNP details by rsID",
+    "database": "snp",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "rsid",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "dbSNP rsID (e.g. rs334, rs7412, rs429358)"
+      }
+    ],
+    "columns": [
+      "rsid",
+      "gene",
+      "chromosome",
+      "position",
+      "alleles",
+      "maf",
+      "clinical",
+      "function"
+    ],
+    "type": "ts",
+    "modulePath": "snp/lookup.js"
+  },
+  {
+    "site": "sra",
+    "name": "download",
+    "description": "Download FASTQ files for an SRA run (via ENA or sra-tools)",
+    "database": "sra",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "accession",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "SRA run accession (e.g. SRR1234567)"
+      },
+      {
+        "name": "outdir",
+        "type": "str",
+        "default": ".",
+        "required": false,
+        "help": "Output directory (default: current directory)"
+      },
+      {
+        "name": "method",
+        "type": "str",
+        "default": "ena",
+        "required": false,
+        "help": "Download method",
+        "choices": [
+          "ena",
+          "sra-tools"
+        ]
+      },
+      {
+        "name": "dry-run",
+        "type": "boolean",
+        "default": false,
+        "required": false,
+        "help": "Show download URLs without downloading"
+      },
+      {
+        "name": "max-size",
+        "type": "str",
+        "required": false,
+        "help": "Max file size to download (e.g. \"500M\", \"2G\"). Larger files are skipped."
+      }
+    ],
+    "columns": [
+      "file",
+      "size",
+      "status"
+    ],
+    "timeout": 600,
+    "type": "ts",
+    "modulePath": "sra/download.js"
+  },
+  {
+    "site": "sra",
+    "name": "run",
+    "description": "Get SRA run details by accession",
+    "database": "sra",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "accession",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "SRA accession (e.g. SRR1234567, SRX1234567)"
+      }
+    ],
+    "columns": [
+      "accession",
+      "title",
+      "platform",
+      "organism",
+      "strategy",
+      "source",
+      "layout",
+      "date"
+    ],
+    "type": "ts",
+    "modulePath": "sra/run.js"
+  },
+  {
+    "site": "sra",
+    "name": "search",
+    "description": "Search SRA sequencing runs",
+    "database": "sra",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "query",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "Search query (e.g. \"RNA-seq human liver\")"
+      },
+      {
+        "name": "limit",
+        "type": "int",
+        "default": 10,
+        "required": false,
+        "help": "Max results (1-200)"
+      }
+    ],
+    "columns": [
+      "accession",
+      "title",
+      "platform",
+      "organism",
+      "samples",
+      "date"
+    ],
+    "type": "ts",
+    "modulePath": "sra/search.js"
+  },
+  {
+    "site": "string",
+    "name": "enrichment",
+    "description": "Functional enrichment analysis for a gene set",
+    "database": "string",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "proteins",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "Comma-separated protein/gene names (e.g. TP53,BRCA1,EGFR,MYC)"
+      },
+      {
+        "name": "species",
+        "type": "int",
+        "default": 9606,
+        "required": false,
+        "help": "NCBI taxonomy ID (default: 9606 human)"
+      }
+    ],
+    "columns": [
+      "category",
+      "term",
+      "description",
+      "fdr",
+      "genes"
+    ],
+    "type": "ts",
+    "modulePath": "string/enrichment.js"
+  },
+  {
+    "site": "string",
+    "name": "network",
+    "description": "Get protein interaction network",
+    "database": "string",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "proteins",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "Comma-separated protein/gene names (e.g. TP53,BRCA1,EGFR)"
+      },
+      {
+        "name": "species",
+        "type": "int",
+        "default": 9606,
+        "required": false,
+        "help": "NCBI taxonomy ID (default: 9606 human)"
+      },
+      {
+        "name": "score",
+        "type": "int",
+        "default": 400,
+        "required": false,
+        "help": "Minimum combined score (0-1000)"
+      }
+    ],
+    "columns": [
+      "proteinA",
+      "proteinB",
+      "score",
+      "experimentalScore",
+      "databaseScore"
+    ],
+    "type": "ts",
+    "modulePath": "string/network.js"
+  },
+  {
+    "site": "string",
+    "name": "partners",
+    "description": "Find protein interaction partners",
+    "database": "string",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "protein",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "Protein/gene name (e.g. TP53)"
+      },
+      {
+        "name": "limit",
+        "type": "int",
+        "default": 10,
+        "required": false,
+        "help": "Max partners (1-50)"
+      },
+      {
+        "name": "species",
+        "type": "int",
+        "default": 9606,
+        "required": false,
+        "help": "NCBI taxonomy ID (default: 9606 human)"
+      },
+      {
+        "name": "score",
+        "type": "int",
+        "default": 400,
+        "required": false,
+        "help": "Minimum combined score (0-1000)"
+      }
+    ],
+    "columns": [
+      "partnerA",
+      "partnerB",
+      "score",
+      "experimentalScore",
+      "databaseScore"
+    ],
+    "type": "ts",
+    "modulePath": "string/partners.js"
+  },
+  {
+    "site": "taxonomy",
+    "name": "lookup",
+    "description": "Look up NCBI Taxonomy by name or ID",
+    "database": "taxonomy",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "query",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "Organism name or Taxonomy ID (e.g. \"Homo sapiens\", 9606)"
+      },
+      {
+        "name": "limit",
+        "type": "int",
+        "default": 5,
+        "required": false,
+        "help": "Max results"
+      }
+    ],
+    "columns": [
+      "taxId",
+      "name",
+      "commonName",
+      "rank",
+      "division",
+      "lineage"
+    ],
+    "type": "ts",
+    "modulePath": "taxonomy/lookup.js"
+  },
+  {
+    "site": "uniprot",
+    "name": "fetch",
+    "description": "Get UniProt protein details by accession",
+    "database": "uniprot",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "accession",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "UniProt accession (e.g. P04637)"
+      }
+    ],
+    "columns": [
+      "accession",
+      "gene",
+      "protein",
+      "organism",
+      "function",
+      "subcellularLocation",
+      "goTerms",
+      "domains"
+    ],
+    "type": "ts",
+    "modulePath": "uniprot/fetch.js"
+  },
+  {
+    "site": "uniprot",
+    "name": "search",
+    "description": "Search UniProt proteins",
+    "database": "uniprot",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "query",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "Search query (e.g. \"TP53\", \"kinase human\")"
+      },
+      {
+        "name": "limit",
+        "type": "int",
+        "default": 10,
+        "required": false,
+        "help": "Max results (1-100)"
+      },
+      {
+        "name": "organism",
+        "type": "str",
+        "default": "9606",
+        "required": false,
+        "help": "Organism taxonomy ID (default: 9606 for human)"
+      },
+      {
+        "name": "reviewed",
+        "type": "str",
+        "default": "true",
+        "required": false,
+        "help": "Only Swiss-Prot reviewed entries (true/false)"
+      }
+    ],
+    "columns": [
+      "accession",
+      "gene",
+      "protein",
+      "organism",
+      "length",
+      "reviewed"
+    ],
+    "type": "ts",
+    "modulePath": "uniprot/search.js"
+  },
+  {
+    "site": "uniprot",
+    "name": "sequence",
+    "description": "Download protein sequence in FASTA format",
+    "database": "uniprot",
+    "strategy": "public",
+    "args": [
+      {
+        "name": "accession",
+        "type": "str",
+        "required": true,
+        "positional": true,
+        "help": "UniProt accession (e.g. P04637) or multiple comma-separated"
+      },
+      {
+        "name": "output",
+        "type": "str",
+        "required": false,
+        "help": "Output file path (default: stdout)"
+      }
+    ],
+    "columns": [
+      "content"
+    ],
+    "type": "ts",
+    "modulePath": "uniprot/sequence.js"
+  }
+]