UC Riverside

Most human common diseases are complex traits that are controlled by genetic variants in multiple genes and their interaction with environmental factors. The rapid evolution of high-throughput sequencing technology has led to tremendous increase in the volume of multi-dimensional omics data with dramatically reduced cost. Advanced bioinformatics tools and quantitative methods are required to understand the molecular and genetic basis of human complex diseases including cancer to advance precision medicine.

In this dissertation, the first chapter introduces the most comprehensive cancer genomic data repository Genomic Data Commons (GDC), the genomic prediction models especially the Best Linear Unbiased Prediction (BLUP) method for common disease risk prediction, and the haplotype phasing methodologies. In the second chapter, a novel R package is developed to download, organize and analyze RNA-seq and miRNA-seq data in GDC to decipher the lncRNA-mRNA related competing endogenous RNAs (ceRNAs) regulatory networks in cancer. In the third chapter, a BLUP-HAT method is proposed to prove the hypotheses that the inclusion of a large number of genes selected from transcriptome and integration of other omic data will greatly improve the predictive power for cancer prognosis. In the fourth chapter, a haplotype phasing method is developed to infer high-resolution chromosome-scale haplotypes using genotype data of a few single gamete cells to facilitate genetic studies of complex traits.