Computational methods and reproducible analysis in regulatory genomics

Monteagudo Mesas, Pablo (2023): Computational methods and reproducible analysis in regulatory genomics. Dissertation, LMU München: Faculty of Chemistry and Pharmacy

Abstract

While the modern field of genomics is exploding with new experimental techniques that push the limits of what is possible, computational methods designed to process and extract useful information from these data try to keep up in order to reliably improve our understanding of gene regulation. In this thesis, we developed reproducible computational pipelines and algorithms to study gene regulation at different levels or stages. We begin by reviewing core ideas required for understanding cells and their regulatory mechanisms and the main experimental sequencing-based assays used to characterize biological systems at the molecular level. We then introduce the two separate projects that comprise this thesis. In our first project, we dissect the contribution of three competing pathways involved in heterochromatic silencing, namely, Pol II occupancy (PO), transcription efficiency (TE), and RNA stability (RS), by comparing heterochromatic and euchromatic regions in Schizosaccharomyces pombe (S.pombe). We characterize each of these regulatory pathways as ratios between expression levels of corresponding high-throughput sequencing assays, PO ( mu ChIP-seq/ wt ChIP-seq), TE (RIP-seq/ChIP-seq) and RS (pA-RNA/RIP-seq), and quantify the relative effects that mutants lacking core components associated with each pathway (i.e., chromatin modifiers, RNAi, and RNA degradation) have on heterochromatic silencing. In our second project, we study how dynamic biological processes, such as development, are regulated and can be characterized at the molecular level by complex (non-linear) single-cell RNA sequencing (scRNA-seq) trajectories, focusing on how such processes can be compared using our novel tool Trajan. We introduce TrajanR, our accompanying R package, that facilitates the standardization and pre-processing of Trajan input data, trajectory inference, and alignment computation under different parameter schemes and provides various visualization options, enabling the analysis of scRNA-seq trajectories in complex settings. We demonstrate the accuracy of Trajan’s alignments through extensive experimentation on simulated data. We also showcase how our TrajanR package facilitates the study of scRNA-seq data based on the analysis of two independent real-world datasets. Finally, we conclude with a discussion of both projects presented in this thesis and an outlook for the future.

• Item Type: Theses (Dissertation, LMU Munich)
• Keywords: RNA-seq, scRNA-seq, Transcriptomics, Computational biology, gene regulation
• Subjects: 500 Natural sciences and mathematics; 500 Natural sciences and mathematics > 540 Chemistry and allied sciences
• Faculties: Faculty of Chemistry and Pharmacy
• Language: English
• Date of oral examination: 1. August 2023
• 1. Referee: Canzar, Stefan
• Signature of the printed copy: 0001/UMC 29856