Differential Co-Abundance Network Analyses for Microbiome Data Adjusted for Clinical Covariates Using Jackknife...

Area of Science:

• Microbiome Research
• Bioinformatics
• Network Analysis

Background:

• Differential network (DN) analysis of microbiome data is crucial for understanding microbial co-abundance across different conditions.
• Existing DN methods often fail to adjust for confounding clinical variables like age and BMI.
• Next-generation sequencing has advanced microbiome analysis, necessitating improved DN techniques.

Approach:

• Propose Statistical Approach via Pseudo-value Information and Estimation for Differential Network Analysis (SOHPIE-DNA), a regression technique using jackknife pseudo-values.
• Incorporate additional covariates (e.g., age, BMI) into the DN analysis framework.
• Demonstrate SOHPIE-DNA's performance through simulations and application to real-world microbiome datasets.

Key Points:

• SOHPIE-DNA achieves higher recall and F1-score compared to existing methods (NetCoMi, MDiNE) in simulations.
• The method maintains comparable precision and accuracy to established techniques.
• SOHPIE-DNA successfully identifies microbial taxa associated with intestinal inflammation and cancer patient fatigue.

Conclusions:

• SOHPIE-DNA offers a robust and adaptable method for differential network analysis in microbiome studies.
• The approach effectively integrates clinical covariates, improving the biological relevance of findings.
• This method advances the understanding of host-microbiome interactions in complex diseases and temporal dynamics.