pvca

pvca quantifies sources of variability, including batch effects, in microarray datasets by combining Principal Component Analysis (PCA) and Variance Components Analysis (VCA).

Key Features:

• Integration of PCA and VCA: Combines Principal Component Analysis (PCA) and Variance Components Analysis (VCA) to assess sources of variability.
• Eigenvalue weighting: Uses eigenvalues associated with PCA eigenvectors as weights to quantify the magnitude of each source of variability.
• Proportional variance expression: Expresses each source's contribution as a proportion of the total variance.
• Mixed-model estimation: Implements variance component estimation via mixed models using the lme4 package in R.
• Random effects and interactions: Fits random effects including two-way interaction terms to model multiple sources of variability.
• Correlation-matrix PCA: Analyzes principal components derived from the original data correlation matrix.
• Batch-effect screening: Serves as a screening method to identify and quantify batch effects before and after normalization in microarray data.

Scientific Applications:

• Batch-effect quantification: Quantifies batch effects and other sources of technical variability in microarray experiments.
• Preprocessing evaluation: Assesses the effectiveness of batch normalization and other preprocessing steps by comparing variance contributions before and after correction.
• Experimental design and integration: Informs decisions about experimental design, data correction, and integration of datasets from multiple batches or studies.

Methodology:

PVCA applies PCA to the original data correlation matrix, fits variance components via mixed models using lme4 in R with random effects and two-way interaction terms, and weights factor contributions by PCA eigenvalues to express each source as a proportion of total variance.