Fig. 3: Variable correlations and feature selection of the antibody functions measured in the study.

a Correlations between the anti-PvDBPII SalI antibody functions are shown. Correlation coefficients (shown in a double red-blue gradient) for each comparison were calculated using Spearman’s rank correlation tests. p values for each significant correlation are indicated, *p < 0.05 **p < 0.01 ***p < 0.001 ****p < 0.0001. b Importance plot for classification of variables specific to PvDBPII SalI that significantly contribute to the PMR reduction using the Boruta algorithm. The importance of each variable is defined as the Z-score of the mean decrease accuracy (normalised permutation importance). Blue boxes correspond to the minimal, average, and maximum Z-scores of shadow features. The variables that contribute significantly (green) or not (red) to the PMR reduction are shown. Boxplots show median Z-score (horizontal bar), interquartile range (boxes), range (whiskers), and outliers (open circles). c Correlations between the PvDBPII PvW1 antibody functions are depicted as in a. d Importance plot for classification of variables specific to PvDBPII PvW1 are depicted as in b.