001, ** = P < 0.01, * = P < 0.05, n/s = P > 0.05. ↑ = Increased in

inflamed vs. non-inflamed tissue, ↓ = Decreased in inflamed vs. non-inflamed tissue. Bold = Phylum level classification, > = Order level classification, >> = Family level classification, >>> = Genus level classification. ∫-LIBSHUFF analysis indicated a significant difference in all of the UC patients and 4 out 6 CD patients. Library Compare analysis confirmed that there were statistically significant differences between inflamed and non-inflamed sites for most of these samples. However, no obvious pattern was apparent and the statistically significant differences were spread between a number of phylogenetic groups (Table 2). Three of the sample pairs that had significant comparisons with ∫-LIBSHUFF (CD3, UC1 and UC5) showed no significant RAD001 in vivo differences with Library Compare. Interestingly, these

discrepancies may be explained by the UniFrac analysis. Unweighted Napabucasin clinical trial UniFrac does not take into account the relative abundances of different phylotypes when comparing communities, only the species overlap. Weighted UniFrac also takes into account the relative abundance of each species. For the three sample pairs with no significant Library Compare results the unweighted UniFrac comparison showed highly significant differences between the paired communities, while the weighted comparison did not (Table 2). This indicates that these paired samples had significantly different community membership Dynein but that the overlapping members of the bacterial community that were present in both samples had similar abundances, thus explaining the significant ∫-LIBSHUFF results and the non-significant Library Compare results. In contrast to this,

the paired set of samples from CD patient 4 were highly significantly different when measured using weighted UniFrac but showed no significance when measured using the unweighted version. Further analysis revealed that a Prevotella species was 3.6 times more abundant in the inflamed than non-inflamed site and accounted for 25% of the total community in the inflamed sample, a difference that was found to be significant to p < 0.00000001 with Library Compare. As the two communities were not recognised as significantly different with ∫-LIBSHUFF and unweighted UniFrac it is possible that this was because, regardless of the differential abundance, overall community membership was similar across both samples. The only sample pair to show no significant differences between inflamed and non-inflamed tissue with either ∫-LIBSHUFF or Library Compare (patient CD6) was characterised by a very low overall diversity, indicating that the microbiota may have been particularly disturbed in this patient.