MaRR

MaRR assesses reproducibility in high-dimensional biological datasets by applying the Maximum Rank Reproducibility (MaRR) nonparametric method to detect reproducible signals, with a particular focus on mass spectrometry-based metabolomics.

Key Features:

• Nonparametric Approach: Does not rely on parametric assumptions about underlying distributions or dependencies of reproducible signals.
• Maximal Rank Statistic: Detects shifts from reproducible to irreproducible signals using a maximal rank statistic suited for high-dimensional data.
• False Discovery Rate Control: Controls the False Discovery Rate (FDR) under conditions typical for MS-Metabolomics, including cases where correlations between replicate pairs diminish for lower-ranked signals.
• High Power and Low Bias: Simulation studies report high power to identify reproducible metabolites and minimal bias, with estimated proportions closely matching true proportions of reproducible signals.

Scientific Applications:

• Reproducibility Assessment: Evaluates consistency of metabolites across technical or biological replicates in MS-Metabolomics studies.
• Multi-center Studies: Applied to multi-center datasets, including the Genetic Epidemiology of Chronic Obstructive Pulmonary Disease (COPD) study, to assess reproducibility across sites and designs.
• Clinical Practice Advancement: Addresses reproducibility challenges in genomic and metabolomic findings to support translation of research results toward clinical applications.

Methodology:

Computes a maximal rank statistic on MS-Metabolomics data across technical or biological replicates using a nonparametric framework, applies FDR control under realistic replicate correlation structures, and has been evaluated via simulation studies assessing power and bias.