VERA SAM

VERA SAM applies a statistical error model to two-color fluorescent DNA microarray intensity data to identify differentially expressed genes using likelihood-based hypothesis testing.

Key Features:

• Statistical error model: Accounts for both multiplicative and additive errors affecting array intensity measurements.
• Parameter estimation (maximum likelihood): Estimates model parameters from observed intensities across all genes via maximum likelihood estimation.
• Generalized Likelihood Ratio Test (GLRT): Performs a GLRT per gene to test for significant differences in dye intensities under the proposed model.
• Error model applications: Uses the error model to improve accuracy of expression ratios and to compare within- and between-slide intensity variations.
• Sample size evaluation: Explores the impact of sample size on parameter optimization and model performance.

Scientific Applications:

• Differential gene expression analysis: Identification of genes with statistically significant expression differences from two-color microarray experiments.
• Comparative expression studies: Comparison of gene expression profiles under different environmental or treatment conditions.
• Yeast galactose response analysis (example): Analysis of gene expression differences between yeast cells grown in galactose-stimulating versus non-stimulating environments.

Methodology:

Employs a statistical model with multiplicative and additive error terms; estimates model parameters by maximum likelihood from observed intensities across all genes; conducts a per-gene GLRT on repeated dye intensity measurements (direct comparisons rather than ratio-based methods) and compares within- and between-slide intensity variations while exploring sample size effects.