arrayMvout

arrayMvout performs multivariate quality assessment of microarray data by applying quality metrics to AffyBatch instances, summarizing those metrics with principal component analysis (PCA), and detecting parametric outliers on principal components while maintaining a fixed Type I error rate.

Key Features:

• Quality Metric Application: Applies multiple quality metrics specifically to AffyBatch instances for assessment of microarray data integrity.
• Dimension Reduction via PCA: Summarizes metric space using principal component analysis to capture major sources of variance.
• Parametric Outlier Detection: Performs parametric multivariate outlier testing on principal components to identify aberrant arrays.
• Fixed Type I Error Rate: Maintains a fixed Type I error rate during outlier identification to control false positives.

Scientific Applications:

• Microarray Quality Assurance: Evaluates and flags low-quality arrays to improve the accuracy and reliability of microarray experiments in human studies.
• Sensitivity and Specificity Analysis: Provides a framework to assess sensitivity and specificity of QA criteria via controlled corruption of intensity data, exemplified by the Affymetrix Latin Square spike-in experiment.
• Enhanced Inferential Power: Improves detection of differential expression in large clinical datasets by excluding arrays identified as outliers, including applications to Affymetrix GeneChips processed from RNA of human peripheral blood samples.

Methodology:

Applies quality metrics to AffyBatch instances, reduces metric dimensionality with PCA, conducts parametric multivariate outlier testing on principal components with a fixed Type I error rate (demonstrated at alpha = 0.01 on Affymetrix and Illumina contributions to the MAQC dataset), and supports controlled corruption of intensity data for method evaluation.