Development, benchmarking, and application of biostatistical methods for population health
Understanding causes of human diseases and disease-related traits is key to prevention and treatment. Germ line genetic data serves an important role, as it cannot be altered by disease processes and can be used to link information from large biological datasets. As biostatisticians, we develop and apply methods to address questions around human health and disease. We have developed methods for colocalisation (do two traits share causal variants?), fine mapping (what are the likely causal variants for a given trait?) and Mendelian randomisation (what is the causal effect of a mediating trait on an outcome trait?) amongst others, and intend to extend these methods in various directions, including to better integrate molecular data from potential mediating traits. This latter data is often from smaller external studies, but the size of UK Biobank and the breadth of its measurements allows us to integrate these external studies with UK Biobank-measured traits.
We have established interests in specific areas of health: immune-mediated, cardiovascular and metabolic diseases, and cancer. We will use our methods, applied to a range of UK Biobank data to identify associated genetic variants and causal mediating traits for these diseases and disease-related traits, and assess their potential as treatment targets, including quantifying the effect of the mediating traits and evaluating what other traits the underlying genetic variants may affect.
We have proposed novel trial design and analysis approaches for patient stratification to predict treatment benefit, and have past and on-going work investigating how to best use historical data for improving efficiency in randomized trials for small populations as well as methods for borrowing information from different subgroups. We would like to inform our future research on these areas, developing methods firmly based on existing genetic data available in UK Biobank.