Fig. 3. Day care children show a distinct and more mature microbial composition in comparison to age-matched home care children.

a ROC curve of random forest result, differentiating between 24 home care and 24 age-matched day care samples, with an AUC of 0.88. b Random forest out of bag (OOB) score for day care and home care samples with a Youden point threshold of 0.51 (see Methods). True positive (TP), true negative (TN), false positive (FP) and false negative (FN) classification results are colored. c A heatmap showing ASVs with significant differential abundance between home care and age-matched day care children samples (paired rank-mean test with dsFDR < 0.1 multiple hypothesis correction, see methods). Each row represents a different ASV (7 ASVs more abundant in day care and 8 more abundant in home care) and each column a different sample. Samples within each day care facility are ordered by age, ASVs are ordered by the effect size, and color bar on the right indicates ASVs taxonomy class. d Venn diagram showing overlap between the 7 day care and 8 home care associated ASVs from panel c (green and red circles respectively) with age younger/older age associated ASVs (blue circle in right and left columns respectively) identified from other cohorts [PRJNA290380²⁰], PRJEB20773¹⁹] - see Methods section for additional details), emphasizing significant larger overlap of home care enriched ASVs with younger subjects (χ² p < 0.05) in contrast to day care enriched ASVs that show a more substantial overlap with older subjects. Source data are provided as a Source Data file.