20Despite the fact that ulcerative colitis (UC) patients show heterogeneous clinical manifestation 21 and diverse response to biological therapies, all UC patients are classified as one group. 22 Therefore, there is a lack of tailored therapies. In order to design these, an unsupervised 23 molecular re-classification of UC patients is evoked. Classical clustering approaches based 24 on tissue transcriptomic data were not able to classify UC patients into subgroups, likely due 25 to associated covariates. In addition, while genome wide association studies (GWAS) have 26 identified potential new target genes, their temporal dynamic revealing the optimal therapeutic 27 window of time remains to be elucidated. To overcome the limitations, we generated time-28 series transcriptome data from a mouse model of colitis, which was then cross-compared with 29 human datasets. This allowed us to visualize IBD-risk gene expression kinetics and reveal 30 that the expression of the majority of IBD-risk genes peak during the inflammatory phase, and 31 not the recovery phase. Moreover, by restricting the analysis to the most differentially 32 expressed genes shared between mouse and human, we were able to cluster UC patients 33 into two subgroups, termed UC1 and UC2. We found that UC1 patients expressed higher 34 copy of genes involved in neutrophil recruitment, activation and degranulation compared to 35 UC2. Of note, we found that over 87% of UC1 patients failed to respond to two of the most 36 widely-used biological therapies for UC. 37This study serves as a proof of concept that cross-species comparison of gene expression 38 profiles enables the temporal annotation of disease-associated gene expression and the 39 stratification of patients as of yet considered molecularly undistinguishable. 40 genes during experimental colitis are yet to be elucidated 9,10 . 61 Furthermore, while there is an obvious clinical heterogeneity among UC patients as seen for 62 example by the location affected (i.e. distal colitis, left-sided and pancolitis, and responder 63 and non-responder) and the extent of the severity, initial treatment for these patient 64 subgroups is identical and modified only if the patients have not responded 6,8 . Biomarkers 65 that could distinguish the different entities of the UC spectrum are currently lacking and they 66 are required in order to achieve the highly needed stratification of UC patients into 67 molecularly functional subgroups 8,11 . Moreover, an unbiased stratification of UC subtypes 68 has never been accomplished at the molecular and functional levels. Here, using 69 transcriptomic data from a well-characterized experimental model of colitis we were able to 70 identify conserved genes between mouse and UC patients. As a result, we were able to gain 71 insights into IBD-risk gene kinetics and to molecularly stratify UC patients in an unsupervised 72 manner. 73 path in PCA space, starting on day 0, passing through day 7 (acute) and ultimately reaching 129 day 14 (recovery) (Fig 2b). Of note, sampl...