Aims

We are interested in developing statistical methods for the analysis of epigenetic data. Specific methods of interest include the use of independent component analysis (ICA) to identify key modes of variation in epigenetic changes through time that might then be correlated with subject exposures. We plan to explore the application of existing ICA algorithms that would operate solely on the epigenetic data and also on new methods that try to use the exposure data to help inform the identification of the latent factors within a joint model. We would also like to consider models that also use genome-wide SNP data to infer latent factors. At the same time the application of these methods may allow artifacts to be identified within the data which could be subsequently removed . This would produce a clean version of the data that could be used by others in any subsequent analysis of this data. Once methods have been developed and tested on simulated data we would apply them to the real data and search for asociations between uncovered latent factors and exposure levels.