{
  "abstract": "We combine GeneNetwork's 1TB structured database with its metadata as an RDF graph and an associated literature corpus. We benchmark frontier and local LLMs querying it through four architectures: document, graph and agentic retrieval-augmented generation (RAG), and a combination of the three. We evaluate them on 40 domain specific expert-curated questions with a new metric: the satisfaction frequency psi (psi, i.e, the share of questions answered with precision and recall of at least 70% or an f1 of at least 60%). psi ranks systems the same way precision, recall and f1 -- a machine learning metric built from the harmonic mean of precision and recall -- but is easier to read and assess for humans. System architecture had a large effect: with full use of Claude Opus 4.8 the plain document RAG of our earlier GNQA system had a psi of 36% against 86% for our new combined tools. Delegating most of the work from Claude Opus 4.8 to a local model halved query costs while still reaching a psi of 59%.",
  "abstract_shown": true,
  "affiliations": [
    "University of Tennessee Health Science Center"
  ],
  "article_type": "New results",
  "arxiv": null,
  "author_total": 8,
  "authors": [
    {
      "name": "J. F. P. Medagbe"
    },
    {
      "name": "B. M. Kilyungi"
    },
    {
      "name": "A. Isaac"
    },
    {
      "name": "A. Kabua"
    },
    {
      "name": "F. Lisso"
    },
    {
      "name": "S. Wambua"
    },
    {
      "name": "R. W. Williams"
    },
    {
      "affs": [
        0
      ],
      "name": "P. Prins"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/8bb902svz3eqm.bib",
  "cited_by": 0,
  "doi": "10.64898/2026.10.04.756532",
  "fulltext_url": "https://www.biorxiv.org/content/10.64898/2026.10.04.756532v1.full",
  "id": "8bb902svz3eqm",
  "issns": [],
  "landing_url": "https://www.biorxiv.org/content/10.64898/2026.10.04.756532v1",
  "language": null,
  "licence": "CC-BY",
  "licence_label": "CC BY",
  "licence_source": "article",
  "licence_url": "https://creativecommons.org/licenses/by/4.0/",
  "metadata_licence": null,
  "native_subjects": [
    [
      "biorxiv",
      "bioinformatics"
    ],
    [
      "openalex",
      "1312|Biomedical Text Mining and Ontologies"
    ]
  ],
  "openalex": "W7221439834",
  "pdf_url": "https://www.biorxiv.org/content/10.64898/2026.10.04.756532v1.full.pdf",
  "pmcid": null,
  "pmid": null,
  "published": "2026-10-04",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/8bb902svz3eqm.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:27:12Z",
      "native_id": "10.64898/2026.10.04.756532",
      "source": "biorxiv",
      "url": "https://www.biorxiv.org/content/10.64898/2026.10.04.756532v1"
    }
  ],
  "subjects": [
    {
      "field": "biology",
      "origin": "biorxiv:bioinformatics",
      "subfield": "bioinformatics"
    }
  ],
  "title": "An Expert-Curated Benchmark for Frontier and Local LLMs Querying a Biomedical Knowledge Graph",
  "title_text": "An Expert-Curated Benchmark for Frontier and Local LLMs Querying a Biomedical Knowledge Graph",
  "type": "preprint",
  "updated_at": "2026-10-30T12:27:12Z",
  "url": "https://research.draconitic.com/paper/8bb902svz3eqm",
  "venue": "bioRxiv",
  "versions": []
}