DiProPerm

DiProPerm performs exact two-sample hypothesis testing for differences between high-dimensional distributions, providing controlled type I error inference for High-Dimensional Low Sample Size (HDLSS) biomedical datasets.

Key Features:

• Exact hypothesis testing: Controls the type I error rate at the nominal level and yields exact p-values for two-sample tests in arbitrary sample-size settings.
• HDLSS applicability: Applicable to high-dimensional, low-sample-size data common in genomics and proteomics where conventional methods may fail.
• Direction-projection approach: Projects data onto directions that maximize differences between groups for focused comparison of distributions.
• Permutation-based inference: Uses permutation techniques to assess the significance of observed differences under the null distribution.
• R package implementation: Implemented as an R package providing functions to perform the DiProPerm test procedures.

Scientific Applications:

• Genomic and proteomic analysis: Testing for distributional differences in high-dimensional genomic and proteomic datasets with limited samples.
• Case-control comparisons: Detecting significant differences between disease and control groups in HDLSS biomedical studies.

Methodology:

Project data onto directions that maximize between-group differences and apply permutation procedures to assess significance, providing exact p-values and controlled type I error.