Fig. 1: Variation in immune responses across individuals.

a A grid showing the assignment of panproteome array datasets (n = 377) to immune clusters. Each row corresponds to an individual (n = 63). Each column corresponds to a sample type (n = 6). The cells are coloured according to the cluster to which the dataset was assigned. b Violin plot showing the birth weights of children assigned to the infant high (n = 23) or infant low (n = 40) immune clusters at 24 mo. c Odds ratios for the associations between different environmental and physiological factors and the assignment of children to the infant high, rather than infant low, immune cluster at 24 mo. Each factor was converted to a binary categorisation of whether individuals (n = 63) belonged to the modal category or not, and a two-tailed Fisher’s exact test used to conduct each univariate analysis. The points represent the maximum likelihood estimates of the odds ratios, and the error bars show the 95% confidence intervals. d Scatterplot showing a t-SNE projection of the variation between samples taken from the umbilical cord, and from the mother at birth. The shape of each point shows the sample type, and the colour of the point shows the cluster to which it was assigned. Each set of samples from an individual are grouped within an ellipse. e Scatterplot showing a t-SNE projection of the variation between samples taken from 12 mo onwards. Data are shown as in (d). Source data are provided as a Source Data file.