T2-DAG

T2-DAG performs high-dimensional two-sample testing to detect differentially expressed gene pathways by incorporating pathway interaction information via graph-informed structural equation modeling.

Key Features:

• Graph-Informed Structural Equation Modeling: Leverages auxiliary gene interaction information from pathway databases to model collective contributions of genes within pathways.
• High-Dimensional Two-Sample Testing: Formulates pathway analysis as a high-dimensional two-sample testing problem to accommodate many genes and limited sample sizes.
• Hotelling's T2-Type Test: Implements an adapted Hotelling's T2-type statistic for detecting differentially expressed pathways.
• Asymptotic Distribution Establishment: Derives the asymptotic distribution of the T2-DAG test under the stated assumptions.
• Robust Performance: Maintains controlled type-I error rates and improved power in simulation studies, including scenarios with incomplete or inaccurate pathway information or unadjusted confounding.

Scientific Applications:

• KEGG Pathway Analysis in Cancer: Detects differentially expressed KEGG pathways across different stages of lung cancer.
• Pathway-Level Genetic Studies: Supports investigation of genetic mutations and their implications for human diseases and traits via pathway-level differential expression.

Methodology:

T2-DAG applies graph-informed structural equation modeling, frames pathway analysis as a high-dimensional two-sample testing problem, implements a Hotelling's T2-type statistic, establishes the test's asymptotic distribution, and assesses performance via simulation studies.