SCOREM: statistical consolidation of redundant expression measures

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• Genome-wide gene expression analysis platforms often use multiple probe sets per gene.
• Discrepancies in expression trends among probe sets for the same gene complicate data interpretation.

Purpose of the Study:

• To develop and validate an algorithm, SCOREM, for assessing probe set agreement in gene expression data.
• To provide methods for consolidating concordant probe sets and analyzing discordant ones.

Main Methods:

• Utilized Kendall's W coefficient of concordance for statistical agreement testing.
• Employed a graph-searching algorithm to identify concordant probe sets.
• Integrated statistical consolidation with sequence analysis for detailed discordant behavior analysis.

Main Results:

• SCOREM effectively determines the level of agreement between probe sets.
• The algorithm can consolidate concordant groups into a single gene value.
• SCOREM identifies biologically meaningful discordant behaviors, such as alternate RNA isoforms and tissue-specific expression patterns.
• Consolidation using SCOREM outperforms other methods in detecting differential expression and overrepresented functional categories.

Conclusions:

• SCOREM offers a robust solution for resolving redundant measures in gene expression data.
• The algorithm enhances the accuracy of genome-wide expression analysis and biological interpretation.
• SCOREM's ability to analyze discordant behaviors provides novel insights into gene regulation.