Fig. 4

Identification of high-confidence plasma ASVs through an in silico decontamination filtering strategy. A Schematic description of the bioinformatics filtering strategy used to identify high-confidence plasma ASVs. *but not exclusively in A, B. B Representative example of an ASV not passing criterion (i) due to a significant differential enrichment in DEB E vs. DEB D (Cupriavidus [blue]) compared to one which passes this criterion where there is no significant abundance difference between DEBs (candidate division OD1 [red]). C A representative example of the filtering strategy for criterion (ii) showing the prevalence in plasma vs. DENC across all ASVs present in DEB E and the decontam [22] classification as ‘real’ or ‘contaminant’. The abundance which is represented by the size of the data points is the average relative abundance (i.e. number of reads normalised by the sample size) of an ASV across plasma samples. D A representative example of an ASV not passing criterion (iii) due to a lack of significant association in the detection between patient-matching plasma samples of DEBs A and B (Comamonadaceae [blue]) compared to an ASV which passes this criterion where there is a significant association between patient-matching samples (Deinococcus [red]). Cohen’s kappa inter-rater reliability coefficient was used to assess the agreement of detection between matching samples. E The prevalence and abundance across plasma samples and DENCs of 31 high-confidence plasma ASVs identified through using filtering criteria (i) to (iv) as described. The abundance represents a log10 transformation of the percentage of reads per plasma sample. F Orthogonal validation of a Faecalibacterium ASV and Clostridium sensu stricto 9 ASV using an ASV-specific ddPCR. Kappa coefficient was used to assess the agreement of detection between the ddPCR and 16S rRNA sequencing results. In B, D and F, the sequence abundance represents the percentage of reads per sample and across the upper and lower panels; each bar corresponds to the same plasma or plasma-DENC sample