iasva

iasva uncovers hidden sources of variation in genomic data using an iterative statistical framework that identifies significant hidden factors for unwanted heterogeneity while adjusting for all known factors.

Key Features:

• Iterative hidden factor identification: Identifies a hidden factor for unwanted heterogeneity while adjusting for all known factors.
• Significance testing: Tests the significance of the putative hidden factor for explaining the unmodeled variation in the data.
• Sequential factor estimation: Uses significant estimated factors as additional known factors in subsequent iterations to uncover further hidden factors.
• Correlated source detection: Uncovers hidden sources of variation even when these sources are correlated.

Scientific Applications:

• Unwanted heterogeneity detection: Identifies hidden factors contributing to unwanted heterogeneity in genomic datasets.
• Batch effect and confounder discovery: Reveals correlated sources of unmodeled variation that may represent technical or biological confounders.

Methodology:

IA-SVA iteratively estimates hidden factors, tests their significance for explaining unmodeled variation, and incorporates significant factors as known covariates in subsequent iterations.