Two-part permutation tests for DNA methylation and microarray data

Area of Science:

• Genomics
• Bioinformatics
• Statistical Genetics

Background:

• Microarray experiments often involve analyzing gene expression data where low or negative values are truncated.
• DNA methylation studies encounter null values representing undetectable methylation alongside observed positive values.
• Standard statistical methods may not adequately handle these data types with truncated or null values.

Purpose of the Study:

• To introduce and evaluate a novel two-part permutation test for analyzing data with truncated or null values.
• To compare the performance of the proposed test against existing two-part tests.

Main Methods:

• Development of a two-part permutation test.
• Application to DNA methylation and microarray data.
• Simulation studies to assess statistical power and performance.

Main Results:

• The proposed two-part permutation test yields smaller p-values compared to the original two-part test for both data types.
• Simulation studies confirm the increased statistical power of the new test.
• The test correctly reduces to a standard test when no truncated or null values are present, without power loss.

Conclusions:

• The two-part permutation test is suitable for routine analysis due to its adaptability.
• It allows for the incorporation of various test statistics and avoids reliance on asymptotic distributions.
• This method is particularly advantageous for microarray analyses with typically small sample sizes.