A strategy for differential abundance analysis of sparse microbiome data with group-wise structured zeros

Area of Science:

• Microbiology
• Bioinformatics
• Statistical Analysis

Background:

• Microbiome analysis using count sequence data faces challenges like zero-inflation, overdispersion, and non-normality.
• Existing methods for differential abundance testing struggle with inflated zero counts and group-wise structured zeros, even after preprocessing.

Purpose of the Study:

• To develop and validate a robust approach for comparing microbial community abundance.
• To specifically address zero-inflation and group-wise structured zeros in microbiome datasets.

Main Methods:

• Proposed a combined approach integrating DESeq2-ZINBWaVE for zero-inflation and DESeq2 for structured zeros.
• Validated the method through extensive simulations.
• Applied the combined approach to two plant microbiome datasets.

Main Results:

• The combined method effectively handles zero-inflation and group-wise structured zeros.
• Successful application to plant microbiome data identified candidate taxa.
• Demonstrated improved performance over existing methods in simulations.

Conclusions:

• The proposed combined differential abundance testing approach enhances the analysis of microbiome count data.
• This method offers a more reliable way to identify differentially abundant taxa in complex microbial communities.
• Facilitates further experimental validation of key microbial taxa in microbiome studies.