RankProd

RankProd identifies differentially expressed genes across multiple microarray studies using a rank product-based meta-analysis.

Key Features:

• Non-Parametric Approach: Employs a non-parametric permutation test to assess significance without assuming a normal distribution of expression values.
• Rank Product Method: Implements the rank product method (Breitling et al.) to combine ranked gene lists across studies as a measure of differential expression.
• Meta-Analysis Capability: Combines results from multiple independent studies to increase statistical power for detecting differentially expressed genes.
• Platform Agnosticism: Accepts pre-processed expression datasets from diverse microarray platforms without relying on platform-specific assumptions.
• False Discovery Rate (FDR): Provides P-values and associated FDR estimates to control for multiple testing errors.
• User-Defined Criteria: Allows specification of user-defined criteria for gene detection and selection.
• Visualization Tools: Produces plots that display actual gene expression levels alongside estimated significance measurements.
• Bioconductor Package: Implemented within the Bioconductor project.

Scientific Applications:

• Genomics and Molecular Biology: Detection of gene expression changes across conditions and studies to support studies in genomics and molecular biology.
• Large-Scale Meta-Analysis: Identification of consistent differential expression patterns across multiple microarray experiments that may be missed by individual studies.

Methodology:

Genes are ranked by expression across datasets, the rank product is computed per gene, and significance is evaluated by permutation testing to produce P-values and FDR estimates.