RummaGEO: Automatic mining of human and mouse gene sets from GEO

Area of Science:

• Bioinformatics
• Genomics
• Computational Biology

Background:

• The Gene Expression Omnibus (GEO) contains millions of samples from numerous studies.
• Current GEO search capabilities are limited to metadata, lacking data-level search functionality.
• There is a need for advanced methods to query gene expression data directly.

Purpose of the Study:

• To develop RummaGEO, a novel gene expression signature search engine.
• To enable data-level searching of human and mouse RNA sequencing perturbation studies within GEO.
• To facilitate the identification of relevant studies based on user-generated gene expression results.

Main Methods:

• Automated identification of sample groups within GEO studies.
• Computation of differential gene expression to extract gene sets.
• Development of a search engine serving gene sets, PubMed, and metadata.
• Analysis of RummaGEO contents for pattern identification and global analyses.

Main Results:

• RummaGEO successfully extracts and serves gene expression signatures from GEO perturbation studies.
• The engine supports searches by gene set, PubMed ID, and study metadata.
• Analysis revealed patterns and enabled global insights from the extracted data.

Conclusions:

• RummaGEO provides a valuable resource for researchers seeking to identify relevant GEO studies using their own gene expression data.
• The tool supports integrative analyses and hypothesis generation within biological research workflows.
• RummaGEO enhances the utility of public gene expression repositories like GEO.