{
  "abstract": "Motivation: Single-cell foundation models (scFMs) are transforming computational biology by enabling generalizable, task-agnostic representations for versatile single-cell analysis. Despite their progress in facilitating rapid deployment for downstream tasks, off-the-shelf scFMs still have some overlooked concerns: (I) (Pretraining Cost.) Pretrain-based scFMs necessitate pretraining on a vast volume of cells, rendering it draining resources in applications. (II) (Heterogeneous Gap.) Large language model (LLM)-based scFMs ignore the tremendous heterogeneous gap between LLM textual and raw cellular spaces, leading to insufficient capability when facing downstream tasks.\n\nResults: To this end, we introduce RAGCell, a versatile single-cell analysis framework that achieves a double-win in both cost-effectiveness and high performance. The success of RAGCell lies in two key aspects: Leveraging LLMs to construct cell-level and feature-level knowledge databases, which serve as supervision signals for training the cell model and significantly reduce the training cost (>pretrain-based scFMs). Aligning cell representations with text embeddings from the bi-level knowledge databases, enabling knowledge transfer from textual spaces to cellular spaces and effectively mitigating the heterogeneous gap (>LLM-based scFMs). Through extensive experiments on five downstream single-cell analysis tasks, we demonstrate that RAGCell achieves outstanding performance compared to state-of-the-art scFMs while operating at less than ∼1/10 the cost of pretrain-based scFMs.\n\nAvailability and implementation: The source code is available at our supplementary file. Embeddings can be found in https://sites.google.com/yale.edu/scelmolib/home",
  "abstract_shown": true,
  "affiliations": [
    "Tsinghua University , Beijing 100084,",
    "Yale University , New Haven, CT 06511,",
    "The Chinese University of Hong Kong , Hong Kong SAR,",
    "The Ohio State University , Columbus, OH 43210,",
    "Nanyang Technological University , Singapore 639798,",
    "Peking University , Beijing 100871,",
    "Tencent Jarvis Lab , Shenzhen 518057,",
    "Westlake University , Hangzhou 310030,"
  ],
  "article_type": null,
  "arxiv": null,
  "author_total": 13,
  "authors": [
    {
      "affs": [
        0,
        1
      ],
      "name": "Tianyu Liu",
      "orcid": "0000-0002-9412-6573"
    },
    {
      "affs": [
        2
      ],
      "name": "Fan Zhang"
    },
    {
      "affs": [
        3
      ],
      "name": "Jiayuan Chen"
    },
    {
      "affs": [
        4
      ],
      "name": "Kun Wang"
    },
    {
      "affs": [
        5
      ],
      "name": "Haoxuan Li"
    },
    {
      "affs": [
        2
      ],
      "name": "Shengju Qian"
    },
    {
      "affs": [
        5
      ],
      "name": "Zhihong Zhu"
    },
    {
      "affs": [
        2
      ],
      "name": "Donghao Zhou"
    },
    {
      "affs": [
        6
      ],
      "name": "Hao Wu"
    },
    {
      "affs": [
        6
      ],
      "name": "Ziheng Zhang"
    },
    {
      "affs": [
        6
      ],
      "name": "Zhenxi Lin"
    },
    {
      "affs": [
        6
      ],
      "name": "Xian Wu"
    },
    {
      "affs": [
        7
      ],
      "name": "Yefeng Zheng"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/792x81xjf297m.bib",
  "cited_by": 0,
  "doi": "10.1093/bioinformatics/btag732",
  "fulltext_url": null,
  "id": "792x81xjf297m",
  "issns": [
    "1367-4803",
    "1367-4811"
  ],
  "landing_url": "https://academic.oup.com/bioinformatics/article/doi/10.1093/bioinformatics/btag732/8860139",
  "language": null,
  "licence": "CC-BY-4.0",
  "licence_label": "CC BY 4.0",
  "licence_source": "article",
  "licence_url": "https://creativecommons.org/licenses/by/4.0/",
  "metadata_licence": null,
  "native_subjects": [
    [
      "openalex",
      "1312|Single-cell and spatial transcriptomics"
    ]
  ],
  "openalex": "W7213291523",
  "pdf_url": "https://academic.oup.com/bioinformatics/advance-article-pdf/doi/10.1093/bioinformatics/btag732/71575345/btag732.pdf",
  "pmcid": null,
  "pmid": null,
  "published": "2026-10-01",
  "publisher": "Oxford University Press (OUP)",
  "ris": "https://research.draconitic.com/paper/792x81xjf297m.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:19:38Z",
      "native_id": "10.1093/bioinformatics/btag732",
      "source": "crossref",
      "url": "https://academic.oup.com/bioinformatics/article/doi/10.1093/bioinformatics/btag732/8860139"
    }
  ],
  "subjects": [
    {
      "field": "biochemistry",
      "origin": "openalex:1312|Single-cell and spatial transcriptomics",
      "subfield": "molecular-biology"
    }
  ],
  "title": "RAGCell: retrieval-augmented generation as supervision for versatile single-cell analysis",
  "title_text": "RAGCell: retrieval-augmented generation as supervision for versatile single-cell analysis",
  "type": "peer-reviewed",
  "updated_at": "2026-10-30T12:19:38Z",
  "url": "https://research.draconitic.com/paper/792x81xjf297m",
  "venue": "Bioinformatics",
  "versions": []
}