High-dimensional test for one-sided hypotheses

Area of Science:

• Statistics
• Bioinformatics
• Genomics

Background:

• High-dimensional data analysis is crucial across scientific domains.
• Existing methods for high-dimensional mean vector testing are often two-sided.
• One-sided tests are needed for detecting directional changes, such as gene up-regulation or down-regulation.

Purpose of the Study:

• To develop and validate a new method for one-sided high-dimensional mean vector testing.
• To address the gap in existing statistical tools for directional analysis in high-dimensional data.
• To apply the new method to a relevant biological problem.

Main Methods:

• Introduction of the Sum Max-Component (SMC) test.
• Asymptotic behavior analysis of the SMC test statistic.
• Extensive validation in finite sample scenarios.
• Application to gene set enrichment analysis using proteomic data.

Main Results:

• The SMC test shows effective performance in finite samples.
• The test achieves competitive rates for type I error and statistical power.
• The method was successfully applied to analyze ovarian cancer proteomic data.

Conclusions:

• The SMC test is a valuable new tool for one-sided high-dimensional mean vector testing.
• The method shows promise for applications in bioinformatics and genomics, particularly in enrichment analysis.
• The study highlights the potential of the SMC test in understanding complex diseases like ovarian cancer.