Information Integration via Bayesian and Neural Network Models With Applications to Biology

Abstract

This thesis is comprised of three parts: 1) we proposed a new method to model convergent rate changes of genomic elements on phylogenetic trees and detect the association between the rate shifts and the convergent phenotypes; 2) we introduced a novel approach for nonparmateric regression using neural networks when the variables have measurement errors; 3) we developed a new method for imputation and clustering for single cell RNA sequencing data.