GSAR

GSAR performs multivariate and aggregation gene set analyses to detect and discriminate alternative hypotheses, including shift versus scale, in differentially expressed gene sets using MST-based non-parametric tests and aggregation statistics.

Key Features:

• Multivariate and Aggregation Tests: Incorporates multivariate tests that account for intergene correlations and aggregation tests that summarize gene-level statistics.
• Minimum-Spanning Tree (MST)-Based Non-Parametric Multivariate Tests: Implements MST-based non-parametric multivariate tests to detect complex multivariate configurations in gene expression data.
• Power and Type I Error Rates: Evaluated via simulation studies showing competitive power and controlled Type I error rates with superior performance in specific parameter spaces.
• Discrimination Against Shift and Scale Alternatives: Distinguishes shift (differential expression) and scale (variance change) alternatives in gene sets.

Scientific Applications:

• Pathway analysis: Applied to pathway analyses to identify pathways exhibiting significant multivariate changes.
• Disease processes: Used to interpret molecular mechanisms in disease processes by distinguishing types of expression changes.
• Drug response studies: Applied to analyze gene expression responses to drugs by discriminating shift versus scale changes.
• Complex biological phenomena: Used for other complex biological phenomena requiring interpretation of multivariate gene set alterations.

Methodology:

Two-step analysis strategy: (1) apply a most-powerful multivariate test to identify pathways where the null hypothesis is rejected; (2) employ MST-based non-parametric multivariate tests on the identified pathways to discriminate specific alternative hypotheses (e.g., shift versus scale).