Integrating transcriptomic and metabolomic data from people with rheumatoid arthritis to predict clinical response to drug treatment

Abstract:

Rheumatoid arthritis is a chronic inflammatory disease characterized by immune-mediated pathology. In the order of 400,000 people in the UK suffer from the disease to varying degrees. The underlying pathology is still not sufficiently understood. However, it is clear that different patients display different responses to treatment with either conventional disease-modifying antirheumatic drugs (cDMARDs) e.g. methotrexate or biologic bDMARDs such as anti-TNF monoclonal antibodies and JAK inhibitors (tsDMARDs). The Scottish Early RA Cohort (SERA n=1173), together with a series of clinical trials that have been led by the University of Glasgow in people with early RA e.g. "ORBIT" and "TASER", have generated a unique biobank within which a range of transcriptomic and metabolomics datasets have been derived but as yet not integrated in terms of data analyses. Specifically, we have RNAseq data available from SERA (early untreated RA and undifferentiated arthritis), RNAseq and mass spec metabolomic data available from TASER (early untreated RA) and RNAseq and shortly metabolomics datasets available from ORBIT (RA failing first cDMARD and randomised to receive rituximab or TNF inhibitor(see McInnes et al. 2017).

Biomarker approaches using pharmacogenomic approaches or using RNAseq alone have been unhelpful so far in RA in offering reliable drug response prediction. Combined metabolomics and trancriptomics data are likely to provide more inference about the factors that distinguish between responders and non-responders, but combination of these data types remains an open problem. This project therefore seeks to address this problem in order to identify new biomarkers arising from combinatorial scoring systems, that can be used to predict likely response and enable choice of the right treatment for each patient.

Aims:

1. Analyse and compare transcriptomic datasets from several RA patient cohorts
2. Analyse and compare metabolomics datasets form several RA patient cohorts
3. Integrate metabolomic and transcriptomic data to seek additional inference on pathways predictive of therapeutic response
4. Generate a panel of metabolite/transcript biomarkers to enable robust prospective prediction of therapeutic outcomes in RA treatment

Training outcomes:

The student will receive directed Bioinformatics training to analyse transcriptomic (Supervisor: Watson) and metabolomics (Supervisor: Barrett) datasets using existing and novel software. They will also receive training in machine learning, Bayesian statistics and computer programming (Supervisor: Rogers) to develop novel algorithms that enable new ways to probe combined metabolomics and transcriptomics datasets. This computational work will feed into clinical understanding of RA (Supervisor: McInnes).

References:

McInnes, I.B., Schett, G. (2017) Pathogenetic insights from the treatment of rheumatoid arthritis. Lancet. 389, 2328-2337. PMID: 28612747

Creek, D.J., Jankevics, A., Burgess, K.E., Breitling, R., Barrett, M.P. (2012) IDEOM: an Excel interface for analysis of LC-MS-based metabolomics data. Bioinformatics. 28, 1048-9. PMID: 22308147

Robert C, Watson M. Errors in RNA-Seq quantification affect genes of relevance to human disease (2015) Genome Biol. 16:177 PMID: 26335491

van der Hooft, J.J., Wandy, J., Barrett, M.P., Burgess, K.E., Rogers, S (2016) Topic modeling for untargeted substructure exploration in metabolomics.. Proc Natl Acad Sci U S A. 113, 13738-13743. PMID: 27856765