Detecting differentially expressed genes in heterogeneous diseases using half Student's t-test.

Area of Science:

• Genomics
• Bioinformatics
• Statistical genetics

Background:

• Microarray technology generates large-scale gene expression data.
• Identifying differentially expressed genes is crucial for disease research.
• Gene expression analysis often involves comparing case and control subjects.

Purpose of the Study:

• To introduce a novel statistical test for identifying differentially expressed genes.
• To address the challenge of gene expression analysis in heterogeneous diseases.
• To improve the sensitivity and specificity of differential gene expression detection.

Main Methods:

• Proposed the 'half Student's t-test' for detecting differential gene expression.
• Utilized Monte Carlo simulations to evaluate the test's performance.
• Compared the proposed test against the conventional 'pooled' Student's t-test.

Main Results:

• The half Student's t-test maintains the nominal alpha level across various distributions.
• Demonstrated superior power compared to the pooled Student's t-test in heterogeneous diseases.
• Showcased significantly higher detection of differentially expressed genes in colon cancer data.

Conclusions:

• The half Student's t-test is recommended for analyzing gene expression in heterogeneous diseases.
• The new test offers improved accuracy and power for identifying disease-related genes.
• This method enhances the ability to discover biomarkers in complex disease studies.