Harnessing genome characterization to uncover disease mechanisms

My project is with the Complex Traits Genetics group, supervised by Dr Sara Knott, Dr Pau Navarro and Prof Chris Haley. The aim of the project is to utilize bioinformatics and statistical analyses alongside existing data on gene annotations and locations of mutations to characterize genetic variants that affect disease phenotypes.

A major aim of precision medicine is to provide personalized treatment for disease, where much of this personalization is based on patient genotype. This relies on a good understanding of genotype-phenotype relations, specifically the causal connection behind certain mutations and their resulting diseases. In an unprecedented era of accurate and abundant genome sequencing, a major challenge is processing the vast amounts of data to provide useful results for clinical targets or drug development. Genome-wide association studies (GWAS) have been tremendously successful for particular diseases in identifying single-nucleotide polymorphisms (SNPs) - base substitutions at a single point in the sequence - that are causally linked to disease phenotypes. However, for most phenotypes we have not uncovered causal mutations, but associated ones, i.e., for most of the associations between SNP variants and phenotype, the specific variant causing the disease and the underlying molecular mechanisms remain unknown. We can supplement GWAS data with gene annotations from genome browsers such as Ensembl, using logistic regression to build multi-variate models that highlight the most influential annotations. Previous members of the group assessed the enrichment and depletion of genetic variants in 54 annotation classes, including genic regions, regulatory features, measures of conservation and patterns of histone modification (see Kindt et al, 2013). This research showed that SNPs associated with the enriched annotations were 8 times more likely to be trait-associated than variants annotated with none of them.
In my PhD project, I aim to further develop these methods to build generalized models for uncovering disease mechanisms. I will find mechanisms that underlie variation across phenotypic categories, including several omics, and those that are different across categories of disease-related phenotypes (such as protein or lipid levels and DNA methylation) and disease (such as diabetes or different types of cancer). I will then conduct further analysis on the most promising findings at trait and region level, using colocalisation and causality analyses, to determine mechanism of causation and control of disease variation.
Kindt et al. (2013) The genomic signature of trait-associated variants. BMC Genomics 14, 108 https://doi.org/10.1186/1471-2164-14-108