Assessment and integration of publicly available SAGE, cDNA microarray, and oligonucleotide microarray expression...

Area of Science:

• Genomics and Bioinformatics
• Molecular Biology

Background:

• Increasing availability of diverse gene expression data (microarrays, SAGE, Affymetrix) necessitates robust validation methods.
• Gene coexpression analysis is a common technique for integrating and validating 'omic' data.

Purpose of the Study:

• To evaluate the utility of publicly available gene expression datasets for global gene coexpression analysis.
• To compare the concordance and reliability of gene coexpression data derived from cDNA microarrays, SAGE libraries, and Affymetrix oligonucleotide microarrays.

Main Methods:

• Analysis of Homo sapiens gene expression data from 1202 cDNA microarray, 242 SAGE, and 667 Affymetrix experiments.
• Assessment of global concordance between datasets using Pearson correlation (rc).
• Validation of coexpressed gene pairs against Gene Ontology (GO) terms for biological process enrichment.

Main Results:

• Low global concordance (rc<0.11) observed between the three analyzed gene expression platforms.
• All platforms identified significantly more coexpressed gene pairs with shared biological processes than expected by chance.
• The Affymetrix platform showed the highest individual performance, with 74% of high-correlation (0.9-1.0) gene pairs confirmed by GO.
• Combining data from multiple expression platforms substantially increased the reliability of coexpressed gene pair identification.

Conclusions:

• Publicly available gene expression datasets show limited global concordance but can identify biologically relevant coexpressed gene pairs.
• Integrating data from diverse expression platforms enhances the reliability and confidence in gene coexpression findings.
• Cross-platform validation against Gene Ontology terms is crucial for assessing the biological significance of coexpressed gene pairs.