Fig. 6

\(W_3\) Profile variations associated to inflammatory status. a Crohn’s disease. Ternary plot in the \(W_1-W_2-W_3\) space of samples colored by disease status (red: Crohn’s disease (CD), green: Non-CD). 95% confidence area of each category are displayed with plain lines. b Boxplot of \(W_1\), \(W_2\) and \(W_3\) levels, structured by disease status. We can observe that CD samples have no marked difference in \(W_1\) levels but are characterized by significantly lower \(W_2\) and strongly higher \(W_3\) levels. This pattern differs from dysbiotic samples where \(W_2\) were over-represented. This observation corroborates the shift of the confidence area in the ternary plot c. c Unraveling dysbiotic and CD profiles. CD and healthy (N) dysbiotic (dys.) and not-dysbiotic(Not_Dys.) samples are displayed in a ternary plot in the \(W_1-W_2-W_3\) space. d Boxplot of the \(W_1\), \(W_2\) and \(W_3\) levels, structured by dysbiotic and CD status. e Mediterranean diet. Ternary plots in the \(W_1-W_2-W_3\) space of samples of the Mediterranean diet cohort. Control and Mediterranean diet groups are displayed in separated ternary plots. For a same individual, samples at baseline (circles) and 8 weeks after intervention (stars) are represented and linked by a line, showing the individual time trajectory. 95% confidence areas are displayed for baseline and 8 weeks groups. Ternary plots are clipped in the \(W_3\) direction at W3 = 0.08. f Boxplots of \(W_3\) levels in the control and Mediterranean diet groups at baseline, 4 weeks and 8 weeks after intervention. \(W_3\) mean and variance are significantly reduced after 8 weeks of Mediterranean diet