Aurora-MCP

{ "title": "All mitochondrial Complex I inhibitors (AI-augmented PubMed mining)", "description": "List of small molecules and drugs claimed to inhibit mitochondrial respiratory Complex I (NADH:ubiquinone oxidoreductase). The compounds were aggregated by automated literature mining of PubMed abstracts using generative AI classification. Each entry includes the compound name, number of PubMed references supporting inhibition, known/new status, a confidence category derived from the reference count, and the list of PubMed IDs supporting the assignment.", "creator": "Daniel Nicorici", "institution": "University of Helsinki, Klefström Lab", "date_created": "2025-10-13", "file_name": "all_mito_complex_I_inhibitors.txt", "delimiter": "TAB", "encoding": "UTF-8", "license": "CC BY 4.0 — attribution required for reuse or redistribution", "fields": { "compound": "Canonical compound or small-molecule name as extracted from PubMed abstracts (case-insensitive).", "pubmed_references": "Integer — number of PubMed abstracts where the compound was identified as inhibiting mitochondrial Complex I (unique PMID count).", "known_status": "Categorical flag: 'known' for compounds established in the literature; 'new' for compounds newly suggested by the AI-based mining pipeline.", "confidence": "Categorical — confidence based on 'pubmed_references' bins: low (≤1), low-medium (2), medium (3–5), high (≥6).", "pubmed_ids": "String — unique supporting PMIDs, sorted ascending and separated by ';'." }, "data_origin": { "sources": [ "PubMed baseline and update XMLs (NCBI)", "AI classification using GPT-4.1-mini (YES/probablyYES → inhibitor)", "Manual curation of high-confidence outputs", "Known inhibitors list (data/reference/mitochondrial_complex_i_inhibitors.txt)" ], "extraction_method": "Generative AI screening of PubMed abstracts mentioning small compounds and Complex I inhibition, followed by aggregation of unique compound names and provenance of supporting PMIDs." }, "intended_use": [ "Reference set for validation of mitochondrial Complex I inhibitor discovery pipelines.", "Training/benchmark data for natural product–mitochondrial function studies.", "Cross-linking with COCONUT, ChEMBL, or Aurora-MCP compound databases." ], "notes": [ "File is TAB-delimited with a single header row.", "Numeric column 'pubmed_references' may be capped at 100 for highly cited compounds (e.g., known-reference boost).", "Entries labeled 'new' have not been manually confirmed in the literature and represent AI-generated predictions.", "'pubmed_ids' contains unique PMIDs, sorted and ';' separated within the cell.", "'confidence' is derived from 'pubmed_references' using fixed bins (low ≤1; low-medium 2; medium 3–5; high ≥6).", "Intended to be used alongside the Aurora-MCP ETL and MCP server environment." ] }