variancePartition

variancePartition quantifies and partitions multiple sources of biological and technical variation in gene expression experiments to attribute variance components across genes and support interpretation of genome-wide transcriptome data.

Key Features:

• Statistical Framework: Employs linear mixed models to quantify the contribution of specified variables to each gene expression trait.
• Comprehensive Analysis: Assesses variation attributable to disease status, sex, cell or tissue type, ancestry, genetic background, experimental stimulus, and technical factors.
• Genome-Wide Summary: Provides genome-wide summaries that prioritize drivers of variation and identify genes that deviate from overall trends.
• Visualization and Interpretation: Includes visualization capabilities to aid interpretation of variance component results.
• Reproducibility Across Datasets: Recovers consistent patterns of biological and technical variation across multiple large-scale transcriptome profiling datasets.

Scientific Applications:

• Disease Biology and Regulatory Genetics: Characterizes sources of variation to inform analyses of disease mechanisms and genetic regulation.
• High-Throughput Genomics Assays: Applicable to gene expression and other high-throughput genomics assays for interpretation of complex datasets.

Methodology:

Uses linear mixed models to decompose variance components associated with different sources for each gene. Applied across multiple large-scale transcriptome profiling datasets to demonstrate reproducibility.