BESC

BESC corrects microarray batch effects by using batch effect signatures (BES) computed from a reference set to predict and remove technical variation while preserving true biological differences.

Key Features:

• Signature-Based Prediction: Uses signatures represented as vectors matching the number of microarray probes, derived from a reference set to capture technical variance while preserving biological signals.
• Blind Correction Capability: Allows precomputation of corrections that can be applied without access to the samples requiring adjustment.
• Individual Sample Adjustment: Applies corrections at the level of individual samples.
• Conservative Approach: Targets removal of known technical effects to avoid altering known and unknown biological differences.
• R implementation: Implemented as an R package.

Scientific Applications:

• Cross-experiment integration: Enables integration of microarray data across different experiments and batches without erasing genuine biological signals.
• Dataset construction and downstream analysis: Facilitates construction of comprehensive datasets from multiple experiments to support more accurate downstream analyses and interpretations.
• Comparative performance: Demonstrated superior performance relative to Surrogate Variable Analysis (SVA), Removing Unwanted Variation (RUV), and Hidden Covariates with Prior (HCP) in maintaining biological differences while correcting technical variance.

Methodology:

BESC estimates individual technical differences using a reference set of microarray samples by calculating batch effect signatures (BES) that capture shared variations and applies these signatures to new datasets to predict and correct batch effects.