Search in the Abstract Database

Abstracts Search 2019

OP10 Systems genomics of ulcerative colitis: combining GWAS and signalling networks for patient stratification and individualised drug targeting in ulcerative colitis

J. Brooks*1,2, D. Modos3, P. Sudhakar4,5, D. Fazekas4,6, A. Zoufir3, A. Watson1,7, M. Tremelling1, B. Verstockt8, S. Vermeire8, A. Bender3, S. Carding2,7, T. Korcsmaros2,4

1Norfolk and Norwich University Hospital, Gastroenterology, Norwich, UK, 2The Quadram Institute Bioscience, Gut Microbes and Health Programs, Norwich, UK, 3Centre for Molecular Science Informatics, Department of Chemistry University of Cambridge, Cambridge, UK, 4Earlham Institute, Norwich Research Park, Norwich, UK, 5KU Leuven,, Department of Chronic Diseases, Metabolism and Ageing, Leuven, Belgium, 6Eötvös Loránd, Department of Genetics, Budapest, Hungary, 7University of East Anglia, Norwich Medical School, Norwich, UK, 8University Hospitals Leuven, Department of Gastroenterology and Hepatology, Leuven, Belgium

Background

The pathogenic signalling pathways of ulcerative colitis (UC) are complex, making patient stratification for optimal therapeutic choices challenging. Disease associated single-nucleotide polymorphisms (SNPs) make the prospect of personalised disease stratification and therapeutics tantalisingly plausible, but forward movement has been difficult. Using systems genomics, we propose a method to identify cohort-specific pathogenic and patient-specific targetable therapeutic pathways.

Methods

Using UC-associated SNPs from the UKIBD Genetics Consortium and the Broad Institute publicly available datasets, we identified the regulatory effects of UC-associated SNPs by identifying those which were localised in transcription factor-binding sites or miRNA target sites. We developed a workflow, iSNP, to identify these regulatory SNPs and build a UC-interactome network (UC-ome). UC-ome contains the regulatory SNP affected genes, the physical interactors of their encoded proteins, the transcription factor and miRNA regulators of the affected gene. The interaction information was integrated from databases containing curated, experimentally validated interactions. We extracted the individual SNP profile from 377 UC patients, and put them through iSNP creating a UC-ome. We used the UC-ome to cluster the patients in cohorts based on their genomic footprint, and compared the clustering with patient-specific clinical parameters with random forest machine learning and enrichment analysis. We validated the workflow on a larger cohort of 941 UC patients from the IBD Biobank in Leuven.

Results

The constructed UC-ome consists of 276 molecules and 1965 physical or regulatory interactions. Analysing the genomic footprints of the patients, we identified 4 patient clusters, and identified common and differing pathogenic pathways between them. We showed that clusters were related to gender and age of onset of disease, but unrelated to therapeutic upscaling of therapy. With machine learning, we identified a subset of patients from within one of the cohorts, for whom the presence of a regulatory MAML2 SNP was a marker for therapeutic upscaling. MAML2 is a Notch pathway activator. This raises the possibility of cohorts of UC patients whose pathway to disease differs to the general UC population, making the MAML2 SNP a potential marker of severity, and the downstream Notch pathway a potential target for personalised therapeutics.

Conclusion

Using a novel systems biology workflow, iSNP, we have been able to analyse the regulatory effects of UC-associated SNPs both on a large cohort level and individual level. We have used this workflow to identify cohorts of patients who may benefit from a therapeutic approach based on their genomic footprint.