All biopsies from non-IBD controls were histologically normal. There was no age difference between CD and UC cases but, due to the indication for colonoscopy, the average age of the non-IBD control patients was higher. The median ages were 32 (25-51) years for the CD group, 26 (24-73) years for the UC group and 51 (45-73) years for the controls. Disease duration was similar. Table 1 Characteristics of patients and biopsy tissue at time of sampling. Diagnosis No. Age Sex Biopsy Site Baron Score Biopsy site Baron

Score CD 1 51 M Rectum 3 Descending 0 CD 2 25 F Descending 2 Descending 0 CD 3 35 F Sigmoid 3 Descending 1 CD 4 29 F Transverse 2 Sigmoid 0 CD 5 35 F Sigmoid 2 Transverse 0 CD 6 26 M Transverse 3 Sigmoid 0 UC 1 49 M Sigmoid 1 Transverse 0 UC 2 26 M Sigmoid 2 Sigmoid 0 UC 3 73 M Rectum 1 Descending BYL719 mw 0 UC 4 25 M Transverse 2 Ascending 0 UC 5 26 M Sigmoid 2 Splenic

0 UC 6 24 F Rectum 2 Descending Alectinib supplier 0 Non-IBD 1 72 F n/a n/a Sigmoid n/a Non-IBD 2 51 F n/a n/a Rectum n/a Non-IBD 3 48 F n/a n/a Rectum n/a Non-IBD 4 45 M n/a n/a Terminal Ileum n/a Non-IBD 5 73 M n/a n/a Descending n/a Quantification of bacterial populations Using qPCR we measured the total bacterial load in the mucosal biopsy samples. The results showed high variability between samples but overall the biopsies from the inflamed intestinal regions of CD patients contained the lowest number of bacteria (Figure 1). The total number of bacteria detected in these inflamed CD samples was significantly lower than the bacterial load present in the inflamed regions of the Decitabine supplier UC patients’ colons. While it appeared

that within each disease cohort the bacterial load was generally lower in inflamed regions of the colon compared to non-inflamed regions the inter-individual variation meant that no other significant differences were detected. Figure 1 qPCR analysis of total bacterial load in mucosal biopsy samples. Figures are mean results for each patient cohort. Error bars denote standard deviation from the mean. Total bacterial load was significantly lower in the inflamed CD biopsies than the UC inflamed biopsies. Overall phylogenetic classification of 16S rRNA gene sequences We next analysed the bacterial diversity in the 29 mucosal biopsy samples by deep sequencing of 16S rRNA gene clone libraries. The final dataset of 10,010 chimera-checked, full-length sequences included an average of 620 clones per CD patient, 750 clones per UC patient and ~350 clones per healthy control. As a whole, the dataset contained an estimated 565 phylotypes (clustered at >99% sequence identity), which could be mapped to eight bacterial phyla. 93% of the sequences belonged to just two of these phyla; the Firmicutes (51.8% of clones) and the Bacteroidetes (41.1%). Within the Firmicutes phylum the vast majority of sequences grouped into two families, the Lachnospiraceae (51.2%) and the Ruminococcaceae (33.