Analysis of, and Software Development for, ChIP-Seq and RNA-Seq Data.

With the recent, rapid progress in high throughput DNA sequencing technology, deep sequencing has made it possible to perform unbiased genome-wide protein-DNA interaction studies (ChIP-Seq) and transcript expression (RNA-Seq) profiling. My thesis focuses on the data analysis and software development for ChIP-Seq and RNA-Seq. The second chapter of my dissertation focuses on applying bioinformatics approaches to integrate sequencing and phylogenetic conservation data to help to predict the genome position of a Gata3 enhancer element that is active exclusively in the T cell lineage. I also performed in vivo imaging to confirm the T cell specificity of the regulatory element. The third chapter consists of a ChIP-Seq study in which I mapped the in vivo binding sites of TR4, an orphan nuclear receptor, in four human ENCODE (ENCyclopedia Of DNA Elements) consortium cell lines. From these data, I discovered that TR4 preferentially binds to and is predicted to regulate genes that play crucial roles in RNA transcription and processing. In the fourth chapter, I performed an analysis of RNA-seq data from differentiating human CD34+ hematopoietic progenitor cells derived from the (fetal) umbilical cord and (adult) bone marrow. Analysis results revealed potential novel isoforms of several known erythroid regulatory proteins and these potential isoforms are supported by the identification of cloned ESTs in the sequence databases. In Chapter 5, I developed a bioinformatics software pipeline to prioritize ChIP-Seq peaks by incorporating external annotation data to facilitate more efficient downstream validation experiments. The developed ChIP-Seq pipeline estimates and adjusts for variation among biological replicates and incorporates peak location relative to gene structure to prioritize resulting peaks. It was then utilized to illustrate its performance with highly localized transcription factor binding (narrow peaks) and more broadly peaked histone modification profiles from ChIP-Seq data. Taken together, we have demonstrated that an integrative approach incorporating greater functional annotation advances our ability to effectively utilize and interpret data from massively parallel sequencing experiments of protein-DNA interactions. Finally, chapter 6 provides an overall summary and conclusions from my research with ChIP-Seq and RNA-Seq data.PhDBioinformaticsUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/94033/1/yuhsuanl_1.pd