Exploring the chromatin landscape and gene expression mechanisms

Abstract

[eng] The interplay between chromatin, transcription factors and genes defines a complex gene regulatory system whose study is essential to understand cell differentiation and how relevant cell functions are maintained or disrupted during biological processes. This is the main area of interest of what is known as 4D genomics, a field that has been evolving on the wave of high-throughput methods producing extensive amounts of data on gene location, chromatin structure and gene expression. 4D genomics data is each day more accessible, the challenge being to interpret it. In this sense Machine Learning (ML) and Artificial Intelligence (AI) methods are becoming crucial to transform noisy experimental data into biological information. The aim of this thesis is to untangle some of the mechanisms that constitute regulatory networks while integrating various state-of-the-art techniques to further understand some of the currently unanswered challenges. During my PhD thesis I have been exploring the chromatin landscape and gene expression mechanisms at different levels of detail and this volume summarizes the main results obtained. In the Introduction, a brief overview of gene expression and transcriptional regulation mechanisms is provided. I also discuss the underpinnings of chromatin organization and stress conditions. Chapters 1 through 5 are a compendium of articles where I describe the different research projects that I undertook during my PhD thesis. These projects have either been published in peer-review journals, are currently under review or in preparation. More specifically in Chapter 1 we started by studying the first regulatory layer, the double-stranded helical structure of DNA and its binding to effector proteins. We developed a ML model that predicted with high accuracy the in vitro affinities and binding sites of various transcription factors based on physical properties of the DNA. Our method also successfully reproduced in vivo data when combined with a second layer of information, the chromatin organization of the nucleosomes. In Chapter 2 we explored nucleosome positioning preferences in yeast genomic DNA by first developing a predictor of nucleosome free regions around the transcription start and terminating sites which are known to comprise critical binding sites. Our method allowed us

to predict the nucleosome architecture within gene bodies by using signal theory from two strongly positioned nucleosomes referred to as +1 and -last (the nucleosomes immediately downstream of the TSS or upstream of the TTS respectively). We additionally studied the link between nucleosome arrangements and gene expression mechanisms. In Chapter 3, the effects of oxidative stress damage on nucleosome organization and overall chromatin structure are described. In order to clarify the effect of these lesions, we performed statistical analysis on a series of gene expression mechanisms through different experimental techniques such as MNase-Seq, Hi-C and Micro-C experiments. Chapters 4 and 5 introduce the study of RNA as a distinct structure and discuss its properties and capability of playing a key role in some regulatory mechanisms such as triplex forming oligonucleotides. A general discussion that encompasses the significance and future perspectives of these 5 projects is presented in Chapter 6, together with the main conclusions of this work in Chapter 7.