Identifying overrepresented concepts in gene lists from literature: a statistical approach based on Poisson mixture...

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Large-scale genomic studies generate extensive gene lists requiring interpretation.
• Current methods rely on manual gene annotation using controlled vocabularies like Gene Ontology (GO).
• Existing vocabularies are often incomplete and manual annotation is labor-intensive, limiting biological domain coverage.

Purpose of the Study:

• To develop a novel statistical method for overrepresentation analysis of gene lists.
• To leverage primary literature (free-text) as a data source for gene set interpretation.
• To address methodological limitations of existing gene list analysis tools.

Main Methods:

• A statistical mixture model framework is employed for overrepresentation analysis.
• The method utilizes free-text from the primary literature as the source of biological concepts.
• Implementation within the BeeSpace literature mining system facilitates interactive gene set analysis.

Main Results:

• The proposed method effectively summarizes important conceptual themes in large gene sets.
• It provides informative results even when traditional GO-based analysis is inadequate.
• Experimental validation demonstrates the program's capability in identifying key biological themes.

Conclusions:

• The developed tool complements existing methods for overrepresentation analysis in genomics.
• It offers biologists a more comprehensive approach to interpreting large gene lists.
• The Genelist Analyzer program is available for use by the research community.