Omics data discovery agents: Agent-supported retrieval, reanalysis, and synthesis of published omics data.

Agent-supported retrieval and reanalysis of published omics data achieved high reproducibility with correlation coefficients of 0.85-0.997 across five studies, demonstrating feasibility.

• Why it matters: Most published omics data remain difficult to access and reuse computationally, limiting the potential for large-scale data synthesis and validation in biomedical research.
• What they did: An agentic framework utilizing large language model agents was developed to fetch, extract, reanalyze, and synthesize omics datasets from thousands of PubMed Central articles, employing containerized analysis tools.
• The result: The system successfully reproduced original data with high accuracy, identified semantically similar studies, and performed meta-analyses, establishing a reusable foundation for automated omics data reuse and validation.