atnick484 / simr2019

The abstract and poster for my internship under the Stanford Institutes of Medical Research at the Khatri Lab in the summer of 2019. More code/scripts can be made available upon request (I used primarily R and SQL)

ABSTRACT

PARTICIPANT: Andrew Nickerson

RESEARCH ADVISORS: Larry Kalesinskas, Purvesh Khatri, PhD

TITLE: Large-scale meta-analysis of chromatin immunoprecipitation-sequenced data to exhibit histone modification relationships

Chromatin immunoprecipitation-sequencing (ChIP-Seq) is a method that allows for the identification of histone modification regions. Sequenced reads contain tags that denote the position of antibody markers, which provide peaks, or sites where histone modifications occur. Large public datasets of ChIP-Seq data exist, provided by large consortia such as ENCODE and Roadmap, which can be leveraged to understand both the epigenetic and transcriptomic basis of disease and cell type. However, to date, ChIP-Seq analysis has largely focused on single experiments and has yet to be examined across experiments.

To have a more systems-level view of biology, we downloaded and processed 2000 ChIP-seq experiments, each of which contained peak data from a single histone modification and cell type in humans. For each peak, we mapped the histone modification to its nearest gene in hg19. We consolidated these data into a well-designed and robust SQL database called ChIP-Map and generated figures for quality control. Having ChIP-Seq data in a database allowed for quick filtering, querying, subsetting, and the application of methods to understand ChIP-Seq data across experiments. To highlight the utility of ChIP-Map, we examined multiple B-cell histone modification experiments and found genes that are highly conserved across histone modifications and genes that are unique to a particular histone modification, building networks, heatmaps, and other figures to examine these relationships. By Venn diagramming certain histone modifications that showed overlap in our heatmap, we found that H3K4me3, H2AFZ, and H3K9ac were mapped near 197 of the same genes. Pathway enrichment analysis revealed many of these genes are involved in mRNA splicing, reaffirming the transcriptional involvement of histone modifications. In the future, we would proceed further with network analysis to identify important histone/gene/cell type relationships as well as develop a web app to allow others to easily explore these data.