Ant Colony Optimisation for the Discovery of Gene-Gene Interactions in Genome-Wide Association Studies

Genome-wide association studies investigate the small changes in DNA among individuals in a population that lead to variations in traits such as height and the propensity to suffer from diseases. Recent advances in genetic technology allow researchers to measure these small differences in DNA in a population (known as single-nucleotide polymorphisms or SNPs) and have already discovered SNPs that are associated with diseases including the widely publicised 'FTO' gene which has been shown to be highly associated with type 2 diabetes. However, single SNPs do not account for all of the variation that is suspected to be inherited and researchers are now beginning to investigate the potential for interactions between multiple SNPs to explain this variation. The number of possible pairs and triplets in the genome though is vast and so a full enumeration search is not possible, meaning that intelligent techniques are required to process the large space of potential interactions. A method that has shown considerable promise in this area is ant colony optimisation (ACO), a nature-inspired search technique based on the way that insects find the shortest path from a nest to a food source in the wild. This search algorithm has two unique properties that make it ideal for this task. The first is that local heuristics can be used to influence the search to find specific gene-gene interactions such as epistasis and the second is that the algorithm creates a pheromone matrix that provides a detailed map of the importance of variables (SNPs) found during the search. This project will investigate the use of ACO to search the space of SNP interactions and their association with a number of diseases including type 2 diabetes and Crohn's disease and also the potential for them to explain human traits such as height. The discovery of these interactions will advance our knowledge of how disease is inherited and could pave the way for highly personalised and pre-emptive treatment based on an individual's genetic makeup.

The chief, longer term (3-10 years) societal impact from this research will be achieved through the identification of gene-gene interactions in genome-wide association studies of human populations and the resulting potential to better stratify individuals suffering from a disease. The ability to accurately determine the disease-risk posed to an individual increases the potential for early and personalised treatment plans. Early or pre-emptive treatment has the potential to delay the onset of symptoms of a disease, reduce symptom severity and complications, and to reduce costs associated with treating the disease across a population of individuals Taking the example of type 2 diabetes, there are an estimated 2.5 million people suffering from this disease currently in the UK and this is expected to rise to 4 million by 2026 (Diabetes UK, 2010). If even a small proportion of these individuals can benefit from diagnosis of disease risk and subsequent early treatment then this would in turn lead to improved health and quality of life for thousands of people and a significant reduction in the burden on the NHS.

The chief, longer term (3-10 years) economic impact will be in the potential to reduce costs of treatments of diseases and their complications through early treatment, an approach that has begun to be used in the USA for type 2 diabetes (Herman et al. 2005). In the UK, the current costs for treating type 2 diabetes are around 10% of the entire budget for the NHS (around £9bn) and this figure is expected to rise significantly (Bagust et al. 2002), so clearly any significant reduction in treatment costs for this disease would have profound economic benefits for the country.

The chief impact on people will be the training of a post-doctoral researcher in this new area of research. The researcher is expected to come to the project with a significant existing skillset, but the precise nature of the algorithms and technologies used in this work means that they will need to become familiar with the particular skills required, rapidly furthering their expertise in this growing field. The proposal will also foster the collaboration between Dr Keedwell and Prof. Frayling, with the attendant increase in knowledge of each field that comes from working together on a project of this novel and interdisciplinary nature.

References
- Bagust, P. K. Hopkinson, L. Maslove and C. J. Currie (2002) "The projected health care burden of Type 2 diabetes in the UK from 2000 to 2060" Diabetic Medicine 19; Supp/4, pp1-5
- Herman et al. (2005) "The Cost-Effectiveness of Lifestyle Modification or Metformin in Preventing Type 2 Diabetes in Adults with Impaired Glucose Tolerance" Ann. Intern. Med. 2005 March 1; 142(5) pp 323-332.