A randomization-based causal inference framework for uncovering environmental exposure effects on human gut...

Area of Science:

• Microbiome research
• Environmental health
• Causal inference statistics

Background:

• Epidemiological cohort studies analyze microbial genomic data to understand environmental impacts on hosts and microbiomes.
• Observational data and complex microbiome characteristics hinder discovering causal links between environment and microbiome.

Purpose of the Study:

• To introduce a causal inference framework based on the Rubin Causal Model for investigating environment-host microbiome relationships.
• To enable testing of plausible sharp null hypotheses and leverage existing statistical tests.
• To uncover potentially causal links between environmental exposure and the gut microbiome using observational data.

Main Methods:

• Developed a causal inference framework using the Rubin Causal Model.
• Applied the framework to the German KORA cohort study data.
• Designed hypothetical randomized experiments (air pollution reduction, smoking prevention) to study gut microbiome effects.

Main Results:

• The framework was illustrated using two hypothetical intervention scenarios.
• Effects on microbial diversity, individual abundances, and network wiring were analyzed.
• In the smoking prevention scenario, a group of taxa including Christensenellaceae and Ruminococcaceae showed potential links to metabolite changes.

Conclusions:

• The proposed statistical framework can uncover potentially causal links between environmental exposures and the gut microbiome from observational data.
• This framework serves as a foundation for future discoveries in environmental health and the gut microbiome's role.