MatchMixeR

MatchMixeR performs cross-platform normalization of gene expression (GE) profiles to remove platform-specific biases while preserving genuine biological variation for integrative analyses.

Key Features:

• Platform-Specific Normalization: Removes platform-specific biases while preserving biological differences using matched GE profiles of identical cell lines or tissues.
• Linear Mixed Effects Regression (LMER): Models and estimates platform differences from matched profiles using LMER to improve the bias–variance trade-off in parameter estimation.
• Computational Efficiency (moment method): Employs a moment method for computationally efficient estimation suitable for ultra-high-dimensional gene expression datasets.
• Concordance and Discovery Balance: Empirical evaluations report superior after-normalization concordance versus competing methods and an improved balance between true and false discoveries in differential expression analyses, notably with limited or unbalanced samples.

Scientific Applications:

• Cross-platform GE integration: Enables combining gene expression datasets generated on different platforms for large-scale integrative studies.
• Differential expression analysis: Supports differential expression analyses on integrated datasets while retaining biological signal across platforms.
• Oncology: Facilitates integrative gene expression studies in oncology to investigate disease mechanisms and therapeutic targets.
• Genomics and personalized medicine: Supports genomics and personalized medicine studies that require aggregation of GE datasets from diverse platforms.

Methodology:

MatchMixeR uses matched GE profiles of the same cell line or tissue measured on different platforms, estimates platform-specific differences using a linear mixed effects regression (LMER) model with computational acceleration via the moment method, and applies the derived model to other datasets to remove platform biases while retaining biological variation.