Legend to Figures:
Figure 1. Airway of severe asthma subjects exhibits distinct species-level microbiome profiles. (A) Barplot showing the average species-level airway microbiome profiles across study cohorts. (B) Boxplot illustrating species-level α-diversity of airway microbiome across study cohorts. (C) Barplots showing log2 fold-change of significantly differentially abundant airway microbial species across study cohorts. (D) Barplots showing log2 fold-change of significantly differentially abundant airway microbial species between the 4 groups of San, SAs/ex, MMA and HC. Significance: ns (not significant,P >0.05), *P ≤0.05, **P ≤0.01 and ***P ≤0.001. Study cohorts: SAn (non-smokers with severe asthma); SAs/ex (smokers or ex-smokers with severe asthma), MMA (non-smokers with mild or moderate asthma) and HC (healthy controls).
Figure 2. Sputum inflammation type shows distinct species-level microbiome profiles.  (A) Barplot showing the average species-level airway microbiome profiles according to granulocytic sputum inflammation type. (B) Boxplot illustrating species-level ⍺-diversity of airway microbiome according to granulocytic sputum inflammation type. (C) Principle coordinate analysis (PCoA) based on Bray-Curtis dissimilarity illustrating species-level β-diversity of airway microbiomes in the 4 granulocytic sputum inflammation types and HC. (D) Barplots showing log2 fold-change of significantly differentially abundant airway microbial species according to granulocytic sputum inflammation type. (E)  Barplots showing log2 fold-change of significantly differentially abundant airway microbial species between each of the sputum granulocytic types and healthy control. Significance: ns (not significant, P >0.05), *P≤ 0.05, **P≤ 0.01 and ***P≤ 0.001. Granulocytic sputum inflammation type: EOS: eosinophilic high (sputum eosinophil % ≥ 1.49%); NEU: neutrophilic high (sputum neutrophil count % ≥ 73.6%);  MIX: both neutrophilic and eosinophilic high; PAU: paucigranulocytic (sputum eosinophil % < 1.49% and sputum neutrophil count % < 73.6%)  and HC (healthy controls).
Figure 3. Molecular phenotypes of asthma exhibit distinct microbiome profiles. (A) Barplot showing the average species-level airway microbiome profiles across the 3 transcriptome-associated molecular phenotypes (TACs) and in healthy controls (HC). (B) Boxplot illustrating species-level α-diversity of airway microbiome across the TACs and HC. (C) Principle coordinate analysis (PCoA) based on Bray-Curtis dissimilarity illustrating species-level β-diversity of airway microbiomes in the 3 TACs and HC (D) Barplots showing log2 fold-change of significantly differentially abundant airway microbial species between the 3 TACs (E) Barplots showing log2 fold-change of significantly differentially abundant airway microbial species between each TAC and HCs. Significance: ns (not significant,P >0.05), *P ≤0.05, **P ≤0.01 and ***P ≤0.001.
Figure 4. Heatmap of Spearman’s correlation coefficientsbetween FEV1 (% predicted), FEV1/FVC ratio, exacerbations in previous year, sputum neutrophil and eosinophil cell counts (%), blood neutrophil and eosinophil counts, and fractional exhaled nitric oxide (FeNO) levels against the abundance of top bacterial species for the 3 asthma cohorts. Significance of Spearman coefficient correlation: ns - not significant (adjusted p > 0.05), with clear box; *adjusted p ≤ 0.05; **adjusted p ≤ 0.01; ***adjusted p ≤ 0.001. SAn: non-smokers with severe asthma; SAs/ex: smokers or ex-smokers with severe asthma; MMA: mild-moderate non-smoking asthma; HC: healthy non-smoking and non-asthmatic.
Figure 5. Airway of severe asthma subjects exhibits temporal stability in species-level microbiome profiles. (A) Barplots showing the baseline and longitudinal follow-up of species-level airway microbiome profiles in subjects with severe asthma, using the same order of subject identifiers. (B) Paired boxplot illustrating species-level α-diversity of airway microbiomes across in severe asthma cases. (C) Principle coordinate analysis (PCoA) based on Bray-Curtis dissimilarity illustrating species-level β-diversity of airway microbiomes at baseline or longitudinal follow-up. Paired samples that are from the same subject are connected by a line. (D) Boxplot showing the species-level microbial compositional similarity across possible sample pairing: Within BL (pairing between any two baseline samples), Within LT (pairing between any two longitudinal follow-up samples), Unpaired (pairing between baseline and longitudinal follow-up samples from different subjects), and paired (pairing between baseline and longitudinal samples from the same subject). Timepoint: BL (baseline) and LT (longitudinal follow-up). Significance: ns (not significant, P >0.05), *P ≤0.05, **P ≤0.01 and ***P ≤0.001.