Comparing Pearson, Spearman and Hoeffding's D measure for gene expression association analysis

Area of Science:

• Genomics
• Bioinformatics
• Systems Biology

Background:

• DNA microarrays are crucial for profiling gene expression.
• Clustering co-expressed genes and constructing regulatory networks require similarity measures.
• Pearson's and Spearman's correlations are commonly used for gene expression analysis.

Purpose of the Study:

• To compare Pearson's correlation, Spearman's correlation, and Hoeffding's D measure for gene expression analysis.
• To assess the most appropriate method for defining associations between gene expression profiles.
• To identify nonlinear and non-monotonic associations in gene expression data.

Main Methods:

• Comparative analysis of three variable association approaches: Pearson's correlation, Spearman's correlation, and Hoeffding's D measure.
• Utilizing simulation studies to evaluate method performance.
• Application of Hoeffding's D measure to identify genes associated with tp53.

Main Results:

• Hoeffding's D measure demonstrates superior performance in identifying nonlinear associations compared to Pearson's and Spearman's correlations.
• Hoeffding's D measure is less sensitive to outliers.
• Hoeffding's D measure effectively identifies nonlinear and non-monotonic associations.

Conclusions:

• Hoeffding's D measure is proposed as a powerful tool for identifying nonlinear associations between gene expression profiles.
• This method enhances the analysis of gene expression data, particularly for complex regulatory relationships.
• Hoeffding's D measure facilitates the discovery of novel gene associations, such as those with tp53.