Developmental dynamics of the epigenome and methods to find relevant regulatory motifs

Machlab, Dania. Developmental dynamics of the epigenome and methods to find relevant regulatory motifs. 2021, Doctoral Thesis, University of Basel, Associated Institution, Faculty of Science.


Official URL: https://edoc.unibas.ch/85652/

Downloads: Statistics Overview


Studying the epigenome and which transcription factors interact with it gives us a better understanding of how developmental processes are regulated and harmoniously orchestrated. For sensory neurons, such signals correspond to environmental stimuli. A group of genes called immediate early genes (IEGs) are known to play important roles during development, and they are some of the first to respond to signals a cell receives. They tend to encode for transcription factors (TFs), are activated within minutes and regulate the activity of other genes. Studying the features of these genes, we found a new epigenetic signature that hints at why they can be induced so fast. They have the active H3K27ac mark on their promoters, and the repressive H3K27me3 mark on their gene bodies. We found a few hundred genes with this signature and called them ‘bipartite’ genes.
Bipartite genes are very lowly expressed, or not at all. They are, however, in a poised state that is even more ready to be quickly induced than the known bivalent genes. The needed transcriptional machinery is already sitting at the promoter. We used t-distributed stochastic neighbor embedding to jointly visualize chromatin accessibility and several histone marks on all genes in barrelette neurons of the somatosensory system in mice. Moreover, we used several developmental time points to visualize genome-wide changes in chromatin states across development. This allowed us to visualize the epigenetic dynamics that bipartite genes undergo by observing how they move from one developmental time point to another in these chromatin landscapes.
As mentioned, IEGs correspond to TFs that have important regulatory roles. Knowing which TFs play relevant or functional roles is key to understanding the underlying developmental processes. Motivated by the importance of finding relevant TFs, we developed computational methods that enable us to make predictions, in an unbiased way, about which TFs could explain an experimental measure of interest, typically coming from sequencing data. We created an R package called monaLisa, short for “motif analysis with Lisa”, that allows for these methods to be used in a user-friendly manner.
The package offers two main ways of identifying regulatory motifs. In the first approach, we made use of an existing method of correcting for sequence composition differences to apply a binned motif enrichment analysis. This method links motif enrichment to an experimental value, for example changes in DNA methylation between two conditions. The second approach uses linear regression to select a set of TFs that are likely to explain the given observations. Specifically, we use randomized lasso stability selection to discover relevant motifs. The new epigenetic signature with the bipartite genes illustrates how the epigenome can control a timely transcriptional response during development, and the methods in monaLisa further enable us to decipher which TFs could be key players.
Advisors:Rijli, Filippo M and Stadler , Michael B and Schübeler, Dirk and Bucher, Philipp
Faculties and Departments:09 Associated Institutions > Friedrich Miescher Institut FMI > Neurobiology > Transcriptional mechanisms of topographic circuit formation (Rijli)
Item Type:Thesis
Thesis Subtype:Doctoral Thesis
Thesis no:14573
Thesis status:Complete
Number of Pages:Verschiedene
Identification Number:
  • urn: urn:nbn:ch:bel-bau-diss145736
edoc DOI:
Last Modified:21 Feb 2022 15:15
Deposited On:21 Feb 2022 15:15

Repository Staff Only: item control page