Berni Canani et al 1 Supplementary Figure 1. Generalized linear model fitting analysis across significantly important 2 taxa (Families as predicted by MetagenomeSeq) across CMA and treatment groups (EHCF and 3 EHCF+LGG). Upper and lower panel represent the GLM model and its ANOVA-based 4 implementation as predicted by ‘rms’ and ‘ResourceSelection’ packages, respectively. ** P < 5 0.00001, * P < 0.001. 6 7 Supplementary Figure 2. Generalized linear model fitting analysis across significantly important 8 taxa (genera as predicted by MetagenomeSeq) across CMA and treatment groups (EHCF and 9 EHCF+ LGG). Upper and lower panel represent the GLM model and its ANOVA-based 10 implementation as predicted by ‘rms’ and ‘ResourceSelection’ packages, respectively. ** P < 11 0.00001, * P < 0.001. 12 13 Supplementary Figure 3. Phylogenetic diversity/evenness and butyrate production across 14 healthy, CMA and treatment groups. (A) box and whisker plot representing alpha diversity i.e. 15 Shannon diversity, across all groups. Color bar intensity represents the butyrate concentration i.e. 16 mmol/kg. (B) Linear regression analysis between phylogenetic evenness and fecal butyrate 17 concentration across healthy CMA and treatment groups. (C) 16S rRNA based beta diversity 18 (phylogeny independent i.e. Bray Curtis) grouping of samples from tolerant and allergic infants 19 Post-EHCF+LGG treatment. 20 1