Genome wide association studies have now reached the scale where one can hope to extract information relevant to clinical applications and public health. The large and complex nature of current datasets will benefit from new and more powerful methods of analysis. We propose to investigate the extent to which a state-of-the-art artificial neural network developed at the Cavendish Astrophysics Group in Cambridge can be applied in this context. We will use the schizophrenia dataset from the Psychiatric Genomics Consortium to study one specific application – the classification of individuals by case or control status. We will compare our results with those from the polygene score which is currently the method of choice for this application.
The goal of this project is to establish whether neural networks can be usefully applied in GWAS analysis given the sample sizes now available.
