{
  "abstract": "Large-scale screening and mapping efforts have produced vast libraries of perturbation-readout data. Converting these measurements into mechanistic insights requires models that link perturbation and genetic input to phenotypes, i.e., labels from experimental readouts, which are usually task specific. We propose a different, virtual experiment modeling approach: train generative models to recreate readouts conditioned on the experimental context, and then let established downstream models extract phenotypes from the synthetic data. As an illustrative case, we develop a bidirectional sequence-image generative framework, CELL-FM, that maps protein sequence and cellular context to fluorescence microscopy images and back, enabling in silico localization prediction, image-conditioned functional motif analysis and generation, and large-scale virtual mutagenesis revealing the amino acid features controlling condensate formation of intrinsically disordered peptides. This approach decouples representation learning from task-specific annotation, reuses rich experimental modalities across many downstream tasks, and preserves the spatial and organizational detail that hand-crafted labels often discard.",
  "abstract_shown": true,
  "affiliations": [
    "University of California, San Francisco"
  ],
  "article_type": "New results",
  "arxiv": null,
  "author_total": 3,
  "authors": [
    {
      "name": "D. Zheng"
    },
    {
      "name": "K. Hong"
    },
    {
      "affs": [
        0
      ],
      "name": "B. Huang"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/7dwvhjndt5h23.bib",
  "cited_by": 0,
  "doi": "10.64898/2026.09.13.751243",
  "fulltext_url": "https://www.biorxiv.org/content/10.64898/2026.09.13.751243v1.full",
  "id": "7dwvhjndt5h23",
  "issns": [],
  "landing_url": "https://www.biorxiv.org/content/10.64898/2026.09.13.751243v1",
  "language": null,
  "licence": "CC-BY-NC",
  "licence_label": "CC BY-NC",
  "licence_source": "article",
  "licence_url": "https://creativecommons.org/licenses/by-nc/4.0/",
  "metadata_licence": null,
  "native_subjects": [
    [
      "biorxiv",
      "bioinformatics"
    ],
    [
      "openalex",
      "1312|Machine Learning in Bioinformatics"
    ]
  ],
  "openalex": "W7213587049",
  "pdf_url": "https://www.biorxiv.org/content/10.64898/2026.09.13.751243v1.full.pdf",
  "pmcid": null,
  "pmid": null,
  "published": "2026-09-13",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/7dwvhjndt5h23.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:23:10Z",
      "native_id": "10.64898/2026.09.13.751243",
      "source": "biorxiv",
      "url": "https://www.biorxiv.org/content/10.64898/2026.09.13.751243v1"
    }
  ],
  "subjects": [
    {
      "field": "biology",
      "origin": "biorxiv:bioinformatics",
      "subfield": "bioinformatics"
    }
  ],
  "title": "Virtual experiments bridge sequence and microscopy with generative models",
  "title_text": "Virtual experiments bridge sequence and microscopy with generative models",
  "type": "preprint",
  "updated_at": "2026-10-30T12:23:10Z",
  "url": "https://research.draconitic.com/paper/7dwvhjndt5h23",
  "venue": "bioRxiv",
  "versions": []
}