Discovering Higher-order SNP Interactions in High-dimensional Genomic Data

Abstract

In this thesis, a multifactor dimensionality reduction based method on associative classification is employed to identify higher-order SNP interactions for enhancing the understanding of the genetic architecture of complex diseases. Further, this thesis explored the application of deep learning techniques by providing new clues into the interaction analysis. The performance of the deep learning method is maximized by unifying deep neural networks with a random forest for achieving reliable interactions in the presence of noise.