Dimensionality of the human phenome

forthcoming

What am I going to do? I will use data from millions of human genomes and thousands of traits (the “phenome”) to quantify the dimension of the heritable human phenome and dissect and map it to single DNA variant resolution; use whole genome sequence data in populations and families to estimate mutational trait heritability; predict future phenotypes using millions of DNA variants and thousands of environmental factors. Why am I going to do this? Despite the discovery of hundreds of thousands of DNA variants that are statistically associated with one or more human complex traits, we do not know which variants are causative, how many traits they affect (pleiotropy), how they function and why there is so much genetic trait variation in the first place. We understand even less of how environmental factors (the “exposome”) interact with genomic variation to cause the variation in phenotypes that we observe. What is missing is a full genome-wide characterisation of multivariate genetic architecture and estimates of genetic variation due to new mutations. A better understanding of pleiotropy, mutational variance and the exposome will inform theoretical and biological models of trait variation and improve prediction of disease in personalised medicine. How am I going to do this? This Project will repurpose data from multiple large biobanks containing millions of genotyped and sequenced individuals who are longitudinally measured on thousands of traits. New advanced statistical approaches will be developed, tested and applied in innovative experimental paradigms that utilise both population and pedigree data. Variance component methods will be developed to estimate mutational variance. Bayesian hierarchical mixture models will be developed and applied to fine-map multivariate trait loci and to predict future disease cases. My Team will develop general user-friendly software tools to achieve the Project aims.