Table 1 The statistical analyses used in this study
Analysis | Data type | Software/package |
|---|---|---|
Box-Cox data transformation was performed to ensure normally distributed data to follow up analysis. The Box-Cox transformation with lambda parameter estimated from data for each variable separately | Raw data | R version 4.0.0 or newer/MASS-package [16] |
Principal coordinate analysis (PCoA) of the Euclidean distances calculated from circulating metabolite values | Box-Cox transformed data | R version 4.0.0 or newer/ape package [17] |
PERMANOVA analysis was used to test whether cohorts’ centroids/mean in the PCoA distance matrix were significantly different from each other. Beta-dispersion (PERMDISP) test was used to examine whether the variance of cohorts was significantly different. ANOSIM test was used to determine whether there is more similarity within the cohorts than between cohorts | PcoA distance matrix | R version 4.0.0 or newer/hagis package [18] |
Hierarchical clustering and heat mapping the Euclidean distance metric and the complete linkage method were used to create clusters based on similarity | The raw metabolite data was scaled column-wise to ensure that metabolite expression values were comparable across samples | R version 4.0.0 or newer/Pheatmap package [19] |
Equality was tested using Levene’s test, and if at least one group showed heteroscedasticity, the ANCOVA test was replaced with a generalized linear model (GLiM) | Box-Cox transformed data | SPSS [20] |
ANCOVA analysis, with covariates (age, sex, BMI), was used to evaluate whether the means of metabolite values were equal or not. The Sidak test was employed for multiple test correction during pairwise comparisons ANCOVA was performed on metabolites that had equal variances between groups | Box-Cox transformed data | SPSS [20] |
A generalized linear model (GLiM) test, with covariates (age, sex, BMI), was used to evaluate whether the means of metabolites values are equal or not. The Sidak test was employed for multiple test correction during pairwise comparisons GLiM test was performed on metabolites that had non-equal variances between groups | Box-Cox transformed data | SPSS [20] |
For data visualization, standardized mean differences (SMD) and SMD 95% confidence intervals were calculated and visualized in forest plot | Box-Cox transformed data | R version 4.0.0 or newer/MBESS-package [21], ggforestplot-package [22] |