Figure 4. Coordinated metabolomic and physiologic responses to dietary perturbation.

Using a linear mixed model for the outcomes of metabolic parameters (e.g., 24-hour energy expenditure, 24-hour respiratory quotient, etc.), we estimated the effect of log₂ fold changes in individual metabolites aggregated across diets. Example mixed model: 24-hour energy expenditure = log₂ fold change metabolite + prechamber log₂metabolite level + diet + age + sex + race + BMI + random intercept per participant. (A) Volcano plots of the β coefficients from the log₂ fold change metabolite variables on respective metabolic parameters. (B) Waterfall plot of the β coefficients on metabolite log₂ fold change for the outcome of postchamber nonesterified fatty acids (NEFA). Model: postchamber NEFA = log₂ fold change metabolite + prechamber log₂ metabolite level + prechamber NEFA + diet + age + sex + race + BMI + random intercept per participant.