Extended Data Fig. 2: Early life abundance of aldh+ bifidobacteria associate with later reduced food allergen–specific IgE in ALADDIN and in the replication cohort PAPS when using 4 time points for faecal abundance annotation between 0 and 6 months and V3 16S rRNA amplicon sequencing but not with reduced food allergen-specific skin prick test in the replication cohort BIS when based on abundance at 2 time points (1 and 6 months) and V4 16S rRNA amplicon sequencing.

a, Maximum relative abundance of aldh+ bifidobacteria across 4 time points (3–6 d, 3w, 2 m, 6 m) in children with or without food allergen–specific IgE (cutoff: 0.35 kUA/L) at 6, until 12, and until 24 months of age in the ALADDIN cohort. b, Maximum relative abundances of aldh+ and aldh− bifidobacteria across 4 time points (w5, 13, 21, 31) in children with or without food allergen–specific IgE (cutoff: 0.35 kUA/L) until 36 months of age in the PAPS cohort. Bacteria were identified using amplicon sequencing of the V3 region of the 16S rRNA gene, and aldh+ and aldh− bifidobacteria were distinguished using two positions that discriminate them in the V3 region. c, Spearman’s rank correlation coefficients (SCC) between the maximum relative abundances of aldh+ bifidobacteria from 0–6 months of life and the maximum concentrations of food allergen–specific IgE in children with food allergen–specific IgE ≥ 0.1 kUA/L at 6 months (n = 43), until 12 months (n = 69), until 24 months (n = 92), and until 60 months (n = 118) in the ALADDIN cohort. The line represents the linear regression line; grey area represents 95% CI around the regression estimate. d, Maximum relative abundances of aldh+ and aldh− bifidobacteria in the BIS cohort between 1 and 6 months of age versus a positive food allergen skin prick test at 12 months of age. Bacteria were identified using amplicon sequencing of the V4 region of the 16S rRNA gene. There is only one discriminative position in the V4 region of the 16S rRNA gene that can be used to distinguish aldh+ from aldh− bifidobacteria, which makes their identification less valid as compared to V3 region identification. a–c, Statistics based on two-sided Wilcoxon rank-sum test. For box plots, the horizontal lines indicate the median; box boundaries indicate the interquartile range; whiskers represent values within 1.5× the interquartile range of the first and third quartiles.