An optimal normalization method for high sparse compositional microbiome data.

Area of Science:

• Bioinformatics
• Computational Biology
• Microbiome Research

Background:

• Omics data, particularly microbiome sequencing, often yield only relative abundance information.
• Existing computational methods to derive absolute microbial abundance from relative data rely on strong assumptions.
• These assumptions limit their applicability to complex study designs, such as multigroup or longitudinal data.

Purpose of the Study:

• To introduce a minimal assumption for converting relative omics data to absolute information.
• To propose the first normalization method operating under this minimal assumption.
• To demonstrate the method's applicability and advantages for multigroup and longitudinal microbiome data analysis.

Main Methods:

• Derivation of a minimal assumption for absolute data extraction from relative omics measurements.
• Development of a novel normalization method based on this minimal assumption.
• Extensive simulation studies to evaluate method performance and compare with existing approaches.

Main Results:

• The proposed method demonstrates optimality and validity under minimal assumptions.
• Existing methods show inconsistent performance when the minimal assumption is not met.
• The new normalization method improves downstream analysis of microbiome data.

Conclusions:

• A novel, assumption-light normalization method enables accurate absolute microbiome abundance estimation.
• This method is robust for complex study designs, outperforming existing techniques.
• The approach facilitates the identification of biologically relevant microbes linked to specific diseases or conditions.