Discovery, refinement and interpretation of genetic variation underlying human complex traits
One of the fundamental challenges in modern human genetics is to convert our continuously improving discovery of disease risk from GWAS studies into an understanding of human biology which is useful for disease management and therapeutic development. Intermediate traits, such as metabolite levels, serum protein levels, physiological measurements and other aspects of normal physiology provide a rich phenotype of natural homeostasis. By mapping the genetic components of variation in these phenotypes we can both understand normal human physiology and explore the relationship with disease processes. This in turn provides useful biomarkers and therapeutic endpoints with clear cut causal relationships with diseases process.
To show the power of this approach we mapped 20 biomedical traits (lipids, hematological, glycemic, inflamatory, renal) in over 35,000 people from 18 studies including the UK10K cohort, replicated in over 100,000 people from 7 studies using whole genome sequencing data and imputation techniques. This has provided an unprecedented perspective on genetic components of normal variation in human physiology. We reconfirmed many previous associations, discovered 17 novel ones and refined the inter-relationship between traits and loci. Given the sample size and the whole genome sequence framework we were further able to fine map 59 loci to credible sets of under 20 variants (e.g. LIPC locus for association with HDL).
Additionally, we developed a robust method, GARFIELD, to associate loci to functional regions of interest (e.g. enhancers, promoters). With its use, we identified traits with ubiquitous enrichment (e.g. Height) and traits with cell-type specific patterns of enrichment (e.g. Crohn’s Disease). Moreover, we show that full blood count GWAS data from the UK Biobank study is enriched in a cell-type specific manner in active enhancers (ChromHMM) from the BLUEPRINT data.
The combination of functional enrichments and intermediate traits provide promising hypotheses of biomarkers and target gene identification.
Date: 4 October 2017, 16:00 (Wednesday, 0th week, Michaelmas 2017)
Venue: Big Data Institute (NDM), Old Road Campus OX3 7LF
Venue Details: BDI LG Seminar Room 0
Speaker: Valentina Iotchkova (MRC Fellow in Computational Biology, Weatherall Institute of Molecular Medicine, University of Oxford)
Organising department: Big Data Institute (NDM)
Organiser: Graham Bagley (University of Oxford, Nuffield Department of Population Health)
Part of: BDI seminars
Booking required?: Not required
Audience: Members of the University only
Editor: Natasha Bowyer