A guideline for the statistical analysis of compositional data in immunology

Area of Science:

• Immunology
• Biostatistics
• Computational Biology

Background:

• Immune cellular composition is crucial in immunology, with large datasets becoming common.
• Immune cell data is compositional, meaning parts sum to a whole, requiring specific statistical handling.
• Standard statistical methods fail to account for correlations inherent in compositional data.

Purpose of the Study:

• To review statistical methods applicable to compositional immune cell data.
• To illustrate regression analyses using log-ratio transformations and Dirichlet regression.
• To apply these methods to immune cell fractions from colorectal cancer patients.

Main Methods:

• Review of statistical methodologies for compositional data analysis.
• Focus on regression analyses employing log-ratio transformations.
• Application of Dirichlet regression as an alternative approach.

Main Results:

• Demonstration of appropriate statistical techniques for immune cellular composition analysis.
• Illustration of log-ratio and Dirichlet regression with real-world immunological data.
• Insights into immune cell fractions in colorectal cancer patients.

Conclusions:

• Compositional data analysis is essential for accurate interpretation of immune cell data.
• Log-ratio and Dirichlet regression provide robust frameworks for such analyses.
• These methods enhance understanding of immune cell dynamics in diseases like colorectal cancer.