A field guide for the compositional analysis of any-omics data

Area of Science:

• Bioinformatics
• Genomics
• Data Analysis

Background:

• Next-generation sequencing (NGS) generates count data for nucleotide abundance.
• Raw NGS counts are influenced by sequencing depth, necessitating normalization.
• Conventional normalization methods rely on assumptions often unmet in heterogeneous samples.

Purpose of the Study:

• To provide a guide for applying compositional data analysis to NGS count data.
• To address limitations of existing normalization techniques.
• To offer a general, assumption-free solution for NGS data analysis.

Main Methods:

• Synthesizing literature on compositional data analysis (CoDa).
• Applying CoDa principles to NGS count data.
• Utilizing log-ratio transformations for data analysis.

Main Results:

• Compositional data analysis offers a general solution for NGS count normalization.
• This approach is assumption-free and valid across diverse datasets.
• Log-ratio transformation is proposed as a key analytical tool.

Conclusions:

• Conventional normalization methods have significant limitations.
• Compositional data analysis provides a superior framework for NGS data.
• Log-ratio transformation enables analysis of relative changes in biological systems.