GEVA

GEVA evaluates variation in differential gene expression across multiple biological conditions by integrating differential expression analysis results and gene expression datasets as an R package.

Key Features:

• Integration of multiple datasets: Accepts and aggregates results from various differential expression analyses to enable cross-study comparison.
• Weighted summarization: Aggregates differential expression signals by assigning weights to different experiments.
• Quantiles partitioning: Divides data into quantile-based segments for downstream analysis.
• Clustering: Groups genes by expression pattern across conditions using clustering methods.
• Factor-based analysis: Categorizes experimental conditions into factors and applies statistical tests including ANOVA (Fisher's and Levene's) to detect factor-specific and factor-dependent effects.
• Gene classification: Classifies genes into Similar, Factor-Dependent, and Factor-Specific categories based on expression variation across conditions.
• Validation and robustness testing: Assessed using 28 transcriptomic datasets across 11 parameter combinations including varied clustering, quantiles, and summarization methods and corroborated with knockout studies.

Scientific Applications:

• Identification of stable versus variable genes: Detects genes with consistent (Similar) or variable (Factor-Dependent, Factor-Specific) expression profiles across conditions.
• Multi-comparison differential expression analysis: Enables comparative evaluation of differential expression across multiple experiments to reveal complex expression dynamics.
• Cross-study integration: Integrates differential expression results from multiple datasets to provide a holistic view of gene expression variability.

Methodology:

Accepts differential expression analysis results as input, applies weighted summarization, quantile partitioning, clustering, performs ANOVA tests (Fisher's and Levene's) for factor-based comparisons, and classifies genes into Similar, Factor-Dependent, and Factor-Specific categories.