Revisiting Fold-Change Calculation: Preference for Median or Geometric Mean over Arithmetic Mean-Based Methods.

Area of Science:

• Biostatistics
• Bioinformatics
• Genomics
• Proteomics
• Metabolomics

Background:

• Fold change is a key metric for analyzing omics data.
• Inconsistent calculation and reporting of fold change introduce discrepancies.
• This study addresses the need for standardized fold change methodologies.

Purpose of the Study:

• To evaluate diverse fold change calculation methods.
• To identify a preferred, robust approach for omics data analysis.
• To enhance the reproducibility of biomedical research findings.

Main Methods:

• Generated artificial datasets with varied distributions (e.g., normal, log-normal).
• Compared fold change calculations against known values in simulated data.
• Analyzed a multi-omics dataset to assess real-world applicability.

Main Results:

• Arithmetic mean-based fold change calculations were frequently inaccurate.
• Inaccuracies were pronounced with differing subgroup distributions or standard deviations.
• Alternative methods (median, geometric mean) demonstrated greater robustness.

Conclusions:

• The arithmetic mean is an inferior method for fold change calculation.
• Median, geometric mean, or paired fold change methods offer improved reliability.
• Standardized, robust fold change calculations and transparent reporting are crucial for accurate interpretation and reproducibility.