Preview

PDF, English Download (17MB) | Terms of use

Abstract

High-throughput technologies are powerful tools for studying fundamental biological processes and for biomedical research. Analysis tools are needed to extract the biological information contained in the generated data sets. In this dissertation I describe methods that I developed for the analysis of data from two high-throughput technologies. The first technology, Circularized Chromosome Conformation Capture (4C), allows to study the 3D chromatin interactions of a certain genomic region (viewpoint) with the rest of the genome. With the second technique, high-throughput microscopy screening, large scale phenotypic screens can be performed to investigate the influences of perturbations, such as compound treatment, on cell phenotypes. Chapter 2 comprises the analysis of three studies which used 4C to study the influence of chromatin 3D structure on gene regulation. The 4C signal shows a strong dependence on the genomic distance from the viewpoint. To address the computational tasks of finding specific interactions, which are superimposed on the regular signal, and to detect changes between interaction profiles of different conditions I developed the R package FourCSeq. The package has been submitted to www.bioconductor.org. Based on my analysis of the interaction profiles from 103 viewpoints, we could show that long-range chromatin interactions were widespread throughout the compact Drosophila genome. Furthermore, the comparison of the interaction profiles from different developmental time points and tissue types revealed that the chromatin configuration was mostly stable across time points and tissue types. In two further 4C sequencing projects, I analyzed the influence of large genomic rearrangements on the chromatin structure of two genomic loci in mice. The first project focused on the locus of the Shh. My analysis showed that upon genomic inversion, the chromatin structure of the locus collapsed and contacts were redistributed. The second project studied the chromatin structure of the Ap2-γ and Bmp7 locus. By analyzing the 4C profiles of 4 viewpoints spaced throughout this locus and determining the primary interaction domains for each viewpoint, I showed that the locus is partitioned by a small transition zone into two distinct domains. Analysis of the interaction profiles from alleles carrying large chromosomal rearrangements further supported this view that the transition zone played an important role in partitioning the locus. Together, both studies show that the chromatin structure is important for long-range gene regulation and allocation of enhancers to their target genes. In Chapter 3, I describe the analysis of a high-throughput microscopy compound screen in a panel of isogenic human colorectal cancer cell lines. In this Pharmacogenetic Phenome Compendium (PGPC) project, we investigated chemical- genetic interactions between compounds and the genetic backgrounds of the isogenic cell line panel. Using high-throughput microscopy, we screened 1280 bioactive compounds in 12 isogenic colon cancer cell lines with specific mutations in signaling pathways. After image segmentation and feature extraction, I used a feature selection algorithm to select a non-redundant set of 20 phenotypic features for further analysis. My analysis of the phenotypic chemical-genetic interaction data allowed us to predict synergistic drug-combinations, uncover connections between signaling pathways, and cluster functionally related compounds and biological processes. Furthermore, I showed that the combined approach of high-throughput microscopy and chemical-genetic screening is more sensitive for interactome mapping than either alone. For easy access to the data and reproducibility of the results, I generated the PGPC R package that will be submitted to www.bioconductor.org.