BatchVaria: variance-based evaluation of batch correction with provenance tracking

BatchVaria reveals that subtype-protected ComBat maintains nearly unchanged subtype contributions after correction, unlike naive ComBat which reduces them by 4.23%.

• Why it matters: Accurate assessment of batch correction methods is crucial for distinguishing true biological signals from technical artifacts, yet current tools lack comprehensive provenance tracking and detailed diagnostics.
• What they did: The authors developed BatchVaria, an R package that integrates correction, variance profiling, and provenance recording within SummarizedExperiment objects, applying it to TCGA breast cancer RNA-seq data across four correction configurations.
• The result: BatchVaria enables detailed comparison of correction methods by analyzing absolute subtype contributions and attribution diagnostics, revealing nuanced differences that variance fractions alone cannot detect, thus improving evaluation of batch correction effectiveness.