VAMPIRE

VAMPIRE performs Bayesian hierarchical analysis of microarray gene expression array data to identify differentially expressed genes by modeling the dependence of measurement variance on expression level.

Key Features:

• Bayesian Hierarchical Modeling: Employs a Bayesian hierarchical model to account for the dependence of measurement variance on gene expression levels, improving variance estimation for low-replicate experiments.
• Variance-Modeled Posterior Inference (VAMPIRE): Implements two Bayesian methods that perform variance modeling followed by integration of posterior probability densities to identify significantly activated or repressed genes.
• Improved Sensitivity and Specificity: Enhances detection sensitivity and specificity for differential expression from microarray data with limited replicates.
• Java Implementation: Provided as Java-based bioinformatics software.
• GOby Integration: Integrates GOby to identify statistically overrepresented gene groups while respecting hierarchical structures such as Gene Ontology (GO).
• Pathway and Transcription Factor Enrichment: Identifies enrichments in Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways and TRANSFAC transcription factor binding sites.

Scientific Applications:

• Genome-wide transcriptional profiling: Detection of differentially expressed genes in genome-wide microarray studies.
• Treatment response analysis: Analysis of transcriptional responses to treatments, exemplified by thiazolidinedione (TZD) effects in diabetic patients.
• Low-replicate microarray experiments: Robust differential expression analysis when experimental replicate numbers are limited.

Methodology:

Modeling gene expression variability using Bayesian hierarchical models; applying two VAMPIRE Bayesian methods involving variance modeling and integration of posterior probability densities for differential expression inference; and using GOby to identify enriched gene groups while respecting GO hierarchy and to test KEGG and TRANSFAC enrichments.