TABLE 2.
The number of metabolites that were significantly different by sex, diet, and PAPOLG genotype and the 2-factor interactions (n = 48)1
| Pathway2 | Sex | Diet | PAPOLG genotype | Sex × diet | Sex × gene | Diet × gene |
| Amino acids (85) | 1123 | 3 | 8 | 1 | 2 | 4 |
| Carbohydrates (23) | 9 | 0 | 3 | 2 | 1 | 0 |
| Cofactors and vitamins (15) | 3 | 1 | 4 | 0 | 1 | 0 |
| Energy (9) | 1 | 0 | 1 | 1 | 0 | 1 |
| Lipids (141) | 15 | 7 | 29 | 6 | 6 | 31 |
| Nucleotides (14) | 1 | 2 | 3 | 0 | 0 | 2 |
| Peptides (14) | 0 | 1 | 0 | 0 | 0 | 0 |
| Xenobiotics (46) | 4 | 3 | 1 | 6 | 0 | 1 |
| Total (347) | 44 | 17 | 50 | 16 | 10 | 39 |
| Median number after Y scrambling3 | 20 | 17 | 20 | 20 | 20 | 20 |
| 4 | <0.001 | NS | <0.001 | NS | <0.001 | <0.001 |
All values are the number of metabolites within each class that reached the achieved statistical threshold (P < 0.05). Details of the metabolites and P values are provided elsewhere (see Supplemental Table 4 under “Supplemental data” in the online issue). ANCOVA was used to analyze differences in metabolites between volunteers. For each metabolite, the postintervention value of the log2 ion intensity was used as the response variable, and the preintervention value was used as a covariate. Sex, diet, and genotype were factor variables, and their main effects and interaction were estimated in the model.
Numbers in parentheses are the total number of metabolites within each class.
The values were derived after 1000 permutations of the data set.
The probability that the total number of metabolites found to be different would have occurred by chance, estimated from the frequency distribution of metabolites that reached the statistical threshold (P < 0.05) after 1000 permutations of the data set.