daMA

daMA implements R/Bioconductor methods for the design and evaluation of efficient microarray experiments on two-colour cDNA and oligonucleotide-based platforms, enabling formulation and optimization of experimental designs to maximize statistical efficiency for targeted contrasts across complex factorial and interaction structures.

Key Features:

• Contrast representation: Represents experimental questions using contrast vectors and contrast matrices to encode targeted comparisons.
• Estimability assessment: Examines the estimability of targeted comparisons under alternative design choices.
• Optimal design theory: Applies principles from optimal design theory to evaluate candidate designs.
• d-efficiency computation: Computes relative d-efficiency of candidate designs for comparing design performance.
• Variance factor comparison: Reduces d-efficiency comparisons to variance factor comparisons for vector-valued contrasts.
• Support for platforms: Targets two-colour cDNA and oligonucleotide-based microarray platforms.
• Factorial and interaction structures: Supports formulation and assessment across simple treatment comparisons and arbitrarily complex factorial and interaction structures.
• Design optimization vs cost: Identifies designs that maximize statistical efficiency while accounting for experimental cost considerations.

Scientific Applications:

• Microarray experiment planning: Planning and optimizing gene expression profiling experiments on two-colour cDNA and oligonucleotide platforms.
• Comparison of candidate designs: Systematic comparison of alternative experimental designs using d-efficiency and variance factors.
• Estimability analysis: Assessing which targeted contrasts are estimable under proposed factorial designs.
• Design selection for complex studies: Selecting efficient designs for experiments with complex factorial and interaction structures.

Methodology:

Represents hypotheses with contrast vectors and matrices, applies optimal design theory to compute relative d-efficiency of candidate designs, and compares variance factors for vector-valued contrasts.