globaltest

globaltest tests groups of genetic variants for association with a response variable, using a penalized regression framework to detect associations including rare variants (mutations present in less than 5% of samples).

Key Features:

• Penalized regression-based global test: Implements a penalized regression approach to assess association between variant groups and phenotypes, improving detection power for rare variants and combined rare and common variant groups.
• Diagnostic plots and multiple testing corrections: Provides diagnostic plots and utilities for multiple testing correction to support result interpretation.
• Gene set testing (GO and KEGG): Performs gene set testing on Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways for pathway-level analysis.
• Variant-specific association information: Reports variant-specific association information within tested groups to inform interpretation of genetic architecture.
• Robustness under low LD: Performs robustly when causal variants exhibit low linkage disequilibrium (LD) between and within causal variants.

Scientific Applications:

• Aggregate-variant association testing: Tests aggregated groups of variants, including rare variants (<5%), for association with phenotypic traits such as simulated hypertension.
• Pathway-level association analysis: Enables exploration of functional implications of genetic associations at the pathway level using GO and KEGG.
• Benchmarking and comparison: Evaluated using Genetic Analysis Workshop 18 (GAW18) data and compared against Score, Sum, SSU, SSUw, UminP, aSPU, aSPUw, the sequence kernel association test (SKAT), and tests using SSU or score statistics with LASSO logistic regression.

Methodology:

Applies a penalized regression framework that integrates penalties into the regression model to manage sparse variant data and enhance association signal detection, particularly under low linkage disequilibrium (LD) conditions.