Extended Data Fig. 2: Effect of hemoglobin genotype and 2-way hierarchical clustering of the full Gouin metabolomic dataset.
From: Metabolome modulation of the host adaptive immunity in human malaria
a–c, Impact of hemoglobin genotype on the serum metabolomes. PCA plots show the clustering patterns of 15 children with non-AA hemoglobin genotype (AC, triangles, n = 10; AS, diamonds, n = 4; and SC stars, n = 1) in principal component analysis of the Gouin children metabolomic dataset before infection (a), during infection (b) and in the combined dataset (c). d, Two-way hierarchical clustering of the full 667 metabolites dataset (rows) in Gouin children (columns) before (n = 99 samples) and during (n = 100) infection. The bar to the left shows children’s infection status (before infection; blue, during infection; red). Clustering was done using Ward’s method. e, Principal variance component analysis of the full Gouin metabolomic dataset showing the proportion of variance of the first three principal components explained by infection status, age, metabolomic experiment run day, sex, hemoglobin level, ratio of lymphocytes to neutrophils and individual effect (left). The bar graph (right) shows the weighted average proportion of variance of the first three principal components explained by each indicated variable.
