Functional response regression model on correlated longitudinal microbiome sequencing data.

Area of Science:

• Microbiome research
• Statistical modeling
• Longitudinal data analysis

Background:

• Functional regression is common for longitudinal data but not established for microbiome sequencing data.
• Existing models assume independent functional responses, limiting their application to correlated microbiome data.
• Analyzing correlated longitudinal microbiome data presents computational challenges.

Purpose of the Study:

• To propose a novel functional response regression model for correlated longitudinal microbiome sequencing data.
• To extend classic functional response regression to handle correlated functional responses.
• To evaluate the relationship between clinical factors and predominant taxa over time in infant gut microbiome data.

Main Methods:

• Developed a functional response regression model for correlated longitudinal microbiome data.
• Derived generalized least squares estimators for correlated functional responses.
• Created a data transformation technique to address computational challenges in analyzing correlated functional response data.

Main Results:

• The proposed method provides unbiased estimations for predictor effects.
• The model demonstrates accurate Type I error and power performance for correlated functional response data.
• Simulations show superior performance compared to the classic functional response regression model.

Conclusions:

• The novel functional regression model effectively analyzes correlated longitudinal microbiome data.
• The method offers a robust approach for microbiome research and understanding temporal dynamics.
• The approach was successfully applied to an infant gut microbiome study, revealing associations between clinical factors and taxa over time.