Fig. 4
Associations between cytokine levels and nasal microbiota at baseline. Using canonical correspondence analysis (CCA) (a) and distance-based redundancy analysis (dbRDA) (b), we assessed the links between cytokine levels day 0 and baseline nasal microbiota. It is assumed that the dependent variables (log10 + 1-transformed relative abundance operational taxonomic units [OTUs]) respond in a unimodal or linear fashion to the predictor variables (log2-transformed cytokine levels) for CCA and dbRDA, respectively. We simultaneously plotted the samples (data points, n = 71), significant (p < 0.05) predictor variables (cytokines; arrows) and the OTUs that were most strongly associated with the first two axes (n = 10 for each axis, excluding overlapping OTUs). Samples were coloured according to carriage3 outcome (red, high-dense carriers, n = 36; blue, non-carriers, n = 18 and orange, low-dense carriers, n = 17), ellipses denote the standard deviation of the samples in each group. Note that carriage3 outcome was not accounted for when simultaneously modelling cytokine/microbiota data, yet still are clearly discriminated, suggesting that baseline microbiota and cytokine levels at day 0 (following live-attenuated influenza vaccine (LAIV) and prior to pneumococcal challenge) are related to pneumococcal receptiveness. Data separation by carriage3 outcome was higher when ordination was based on both microbiota and cytokine data (dbRDA and CCA; standardized absolute β-coefficient 0.43 and 0.32, respectively) compared to microbiota alone (non-metric multidimensional scaling [NMDS]; 0.22; Fig. 2). This was also true when coefficients were split between X- or Y-coordinates. An extensive description on our method to compare data separation by carriage3 outcome is provided in the Supplementary Methods section. CAP, constrained analysis of principal coordinates; GM-CSF, granulocyte–macrophage colony-stimulating factor; VEGF, vascular endothelial growth factor; Cor, Corynebacterium; Dol, Dolosigranulum; Pep, Peptoniphilus; Mor, Moraxella; Fin, Finegoldia; Str, Streptococcus; Hae, Haemophilus; Fus, Fusobacterium; Ana, Anaerococcus; Act, Actinobacillus; Pre, Prevotella; All, Alloprevotella; Por, Porphyromonas; Vei, Veillonella; Agg, Aggregatibacter and Sel, Selomonas. Numbers correspond with overall mean relative abundance rank
