{
  "abstract": "Protein language models have become a dominant foundation for sequence-based protein-protein interaction (PPI) prediction, but their generalization remains limited under stringent evaluation, and benchmark accuracy alone cannot reveal whether a PPI predictor learns partner-specific biological signals or exploits contextual shortcuts. Here we introduce AuditPPI, an interpretable framework that transforms sparse-autoencoder (SAE) features from a frozen protein language model into order-invariant pair fingerprints, recasting PPI prediction as an auditable tabular-learning problem. AuditPPI achieves competitive predictive performance across diverse PPI benchmarks while enabling multiscale auditing of the information supporting its predictions. Protein-level audits show that partner-independent participation signals remain substantial in conventional benchmarks, including protein-disjoint benchmark, but are largely insufficient for pair-level discrimination in topology- and degree-controlled benchmark. Pair-level analyses further show that removing protein identity overlap does not eliminate contextual structure: protein-disjoint benchmarking retains strong subcellular co-localization signals and benchmark-dependent feature matching captured by SAE co-activation and absolute-difference. Structural analyses, however, provide little evidence that these predictive signals are specifically grounded in PPI interfaces: feature enrichment at interfaces for protein-disjoint benchmark was no longer detectable after controlling for surface exposure, and contact-specific enrichment was observed for only a small fraction of testable SAE feature pairs. AuditPPI therefore separates predictive success from mechanistic fidelity and provides a practical framework for evaluating whether sequence-based PPI predictors capture partner-specific biological signals instead of contextual shortcuts.",
  "abstract_shown": true,
  "affiliations": [
    "Shanghai Jiao Tong University"
  ],
  "article_type": "New results",
  "arxiv": null,
  "author_total": 7,
  "authors": [
    {
      "affs": [
        0
      ],
      "name": "W. Zhu"
    },
    {
      "name": "S. Wang"
    },
    {
      "name": "X. Liu"
    },
    {
      "name": "Y. Xue"
    },
    {
      "name": "H.-B. Shen"
    },
    {
      "name": "B. Chen"
    },
    {
      "name": "X. Pan"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/e3nw1eygqyezy.bib",
  "cited_by": 0,
  "doi": "10.64898/2026.09.27.754758",
  "fulltext_url": "https://www.biorxiv.org/content/10.64898/2026.09.27.754758v1.full",
  "id": "e3nw1eygqyezy",
  "issns": [],
  "landing_url": "https://www.biorxiv.org/content/10.64898/2026.09.27.754758v1",
  "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|Machine Learning in Bioinformatics"
    ]
  ],
  "openalex": "W7216499603",
  "pdf_url": "https://www.biorxiv.org/content/10.64898/2026.09.27.754758v1.full.pdf",
  "pmcid": null,
  "pmid": null,
  "published": "2026-09-27",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/e3nw1eygqyezy.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:25:49Z",
      "native_id": "10.64898/2026.09.27.754758",
      "source": "biorxiv",
      "url": "https://www.biorxiv.org/content/10.64898/2026.09.27.754758v1"
    }
  ],
  "subjects": [
    {
      "field": "biology",
      "origin": "biorxiv:bioinformatics",
      "subfield": "bioinformatics"
    }
  ],
  "title": "Auditing Protein-Protein Interaction Signals with Sparse Autoencoder Fingerprints",
  "title_text": "Auditing Protein-Protein Interaction Signals with Sparse Autoencoder Fingerprints",
  "type": "preprint",
  "updated_at": "2026-10-30T12:25:49Z",
  "url": "https://research.draconitic.com/paper/e3nw1eygqyezy",
  "venue": "bioRxiv",
  "versions": []
}