Defining novel biomarkers and therapeutic targets for type 2 diabetes and obesity through genetic analysis in large-scale human data sets.

Background: T2D and obesity are major contributors to ill health globally. There is a critical need for novel preventative and therapeutic strategies against these diseases supported by robust stratification of individual risk and disease subtype. Progress to date has been hampered by poor understanding of the mechanisms responsible for disease development and progression. Human genetics provides a powerful means to highlight pathways of relevance to human disease, and there has been a spectacular increase in the interest in using such information to guide drug and biomarker development. This proposal seeks to build on existing and ongoing efforts at risk-variant discovery led by the supervisors and to advance understanding of disease mechanisms in ways that have direct translational relevance. It will test the hypothesis that the integration of diverse types of large-scale human genetic and genomic data, within a “mendelian randomisation” framework, represents a powerful strategy for the identification and characterisation of novel pathways causal for disease, and that such discoveries will have particular translational value in terms of providing validated biomarkers and therapeutic targets.

Description of the work: The primary focus of the research to be performed will be computational and statistical and the main activities will involve:
•integration of diverse human genetics GWAS data-sets, including UKBiobank, to generate data sets of >1 million individuals, accelerating existing efforts in global T2D and obesity consortia
•analysis of molecular phenotype data (proprietary and public) available to the supervisors, and gathered from tissues of particular relevance to T2D and obesity (including RNA expression data from human pancreatic islets, subcutaneous fat and skeletal muscle as well as serum proteomic and metabolomics data), the objective being to characterise the “multivariant” predictors of those phenotypes; and
•application of two sample mendelian randomisation (“2SMR”) methods to link these two data types.

The objective will be to identify molecular phenotypes “causal” for disease: in other words, those that are responsible for mediating the action of genetic (and possibly non-genetic) risk factors on disease development and progression. Such biomarkers have translational potential with respect to (a) stratification of individual disease risk; (b) stratification of disease subtype; (c) monitoring need for and response to intervention; and (d) prioritisation of validated therapeutic targets for drug development. A particularly relevant example of the clinical impact of such a molecular biomarker is the use of cholesterol in relation to vascular disease.