91 ± 1.56 <0.0001 23.97 ± 1.36 0.9945 29.39 ± 1.51 Subject 2 55.64 ± 1.51 <0.0001 27.31 ± 1.41 0.9849 31.78 ± 1.44 Subject 3 23.86 ± 1.37 <0.0001 10.27 ± 0.97 0.1584 8.99 ± 0.89 Subject 4 38.60 ± 1.53 <0.0001 16.05 ± 1.19 0.6741 16.83 ± 1.17 SGII           Subject 1 48.13 ± 1.61

<0.0001 28.50 ± 1.40 0.9947 34.07 ± 1.56 Subject 2 50.75 ± 1.55 <0.0001 21.64 ± 1.31 0.2537 20.50 ± 1.25 Subject 3 35.31 ± 1.51 <0.0001 7.64 ± 0.84 0.9827 3-deazaneplanocin A 10.37 ± 0.99 Subject 4 52.52 ± 1.57 <0.0001 25.78 ± 1.39 0.9439 28.95 ± 1.41 aBased on the mean of 10,000 iterations. 1,000 random spacers were sampled per iteration. bEmpirical p-value based on the fraction of times the estimated percent shared spacers for comparisons within skin or saliva exceeds that between skin and saliva. p-values ≤0.05 are represented in bold. We also examined CRISPR repertoires by collapsing all time points between subjects to determine whether the CRISPR spacers in each environment were a direct reflection of the subject and environment from which they were derived. When considering both the presence of spacers and their abundance in skin and saliva, we found Cetuximab that for most subjects the CRISPR repertoires were significantly subject-specific (Additional file 1: Table S5). We estimated that 94% of the SGII spacers were conserved across

the skin and saliva of Subject #1 compared to only 35% when comparing between different subjects (p < 0.0001). Similar results were produced for all subjects ioxilan for both SGI and SGII CRISPR spacers with the exception of Subject #4 (Additional file 1: Table S5). While the results did not reach statistical significance for Subject#4, the trends in the proportions of intra-subject shared spacers between skin and saliva exceeded inter-subject comparisons substantially

(86% vs 57% for SGI spacers and 58% vs 35% for SGII spacers). CRISPR spacer matches We tested whether the spacer repertoires from skin and saliva matched similar viruses (Additional file 2: Figure S6). We found that 8.6% of saliva-derived and 25.3% of skin-derived SGII spacers were homologous to streptococcal viruses in the NCBI Non-redundant (NR) database, and 6.9% of saliva-derived and 15.3% of skin-derived SGI spacers were homologous to streptococcal viruses. Comparatively, only 4.5% of saliva-derived and 6.5% of skin-derived SGII spacers were homologous to streptococcal plasmids, and 0.3% of saliva-derived and 0.9% of skin-derived SGI spacers were homologous to streptococcal plasmids. In all cases, the proportion of skin-derived spacers with homologues in the NR database was significantly (p ≤ 0.005) greater than that for saliva-derived spacers. We created heatmaps of the spacer homologues across all time points for both saliva and skin, where only spacers that were newly identified at each time point were included.