Understanding sequencing data as compositions: an outlook and review

Area of Science:

• Genomics
• Bioinformatics
• Statistical Analysis

Background:

• Sequencing count data are inherently compositional due to fixed library sizes.
• This compositional nature violates assumptions of standard statistical methods.
• Uncorrected analyses can lead to erroneous biological interpretations.

Purpose of the Study:

• To review compositional data analysis (CoDA) principles.
• To demonstrate why sequencing data are compositional.
• To discuss valid analytical methods for sequencing data.

Main Methods:

• Review of compositional data analysis (CoDA) principles.
• Explanation of the compositional nature of sequencing data.
• Discussion of compositionally valid statistical methods.

Main Results:

• Sequencing data are confirmed as compositional, requiring specialized analysis.
• Compositional data analysis (CoDA) offers valid approaches for sequencing data.
• Standard statistical methods are inappropriate for raw sequencing counts.

Conclusions:

• Accurate interpretation of sequencing data necessitates understanding its compositional nature.
• Compositional data analysis (CoDA) provides the appropriate framework for analyzing sequencing data.
• Adoption of CoDA methods will improve the reliability of genomic studies.