Statistical ‘omics approaches to understanding autoimmune diseases

Genetic association studies have identified many positions on the human genome where a change in DNA sequence influences the risk of autoimmune diseases such as juvenile idiopathic arthritis and type 1 diabetes. However, we do not often understand how the sequence change affects disease risk: through which gene, in which cells, and how these risk variants combine in biological pathways. I will use statistical techniques to combine information on genetic association with ten different autoimmune diseases and genetic association of cell specific gene expression variation to identify causal genes and cells in these diseases. I will use the joint disease association datasets to identify and partition disease risk across the cell specific pathways within which these disease causal genes act, and look for partitions of patients corresponding to greater risk in some subset of these pathways which also correlates with clinical or phenotypic data. The existence of such partitions will identify opportunities for genetically supported stratified medicine, targeting the right drug to the right patient according to phenotypes measurable in the clinic.

Autoimmune diseases collectively affect 5% of the population, and clinical presentation of one disease alters risk of another, consistent with the reported overlaps between genetic association profiles of autoimmune diseases. Diagnosis frequently leads to lifelong chronic ill health and reduced lifespan while treatment options are limited. Identification of new pharmaceutical targets, such as the biological pathways through which disease risk is modified by genetic variation is a priority. Identification of genetically validated partitions of patients within a disease according to relevant risk pathways could enable stratified medicine: delivering the right treatment that has been mechanistically matched to the right subset of patients.
GWAS studies have identified hundreds of associations of diseases with genetic variation, but translation to biologicial understanding and clinical impact has been slow. This programme aims develop and apply the integrative statistical methods needed to genomic data and immunological data to better understand the causes of the diseases and the heterogeneity between and within different clinical diagnoses.
Research objectives are:
1. develop methods for fine mapping candidate causal genetic variants in full genotype or summary GWAS data that are not limited by assumptions of independence between causal variants and that leverage information across diseases;
2. develop statistical tools to integrate GWAS, eQTL, and promoter targeted, high-throughput chromosome conformation capture (Hi-C) datasets from primary immune cells to define candidate causal genes, cells and pathways for autoimmune disease;
3. use longitudinal gene expression data from humans subject to some immune stimulation (pharmaceutical intervention or infection) to define specific networks (pathways) of genes involved in immune responses;
4. identify which of these pathways are enriched for autoimmune disease-associated variation and compare the enrichment between diseases;
5. partition genetic risk loci into pathway-defined immune disease gene sets;
These methods will be tested on simulated data and applied to genetic association data on ~100,000 subjects and smaller detailed cellular phenotyping dataset available publicly or through collaborations with the autoimmune disease research community in the UK and internationally.