Development of bioinformatics methods to investigate characteristics of transcription factor binding during early embryonic development and cell differentiation

Abstract

Epigenetic mechanisms drive determination of cell types from embryonic development to differentiated adult tissues. Transcription factors are particularly important for these processes due to their ability to bind specific DNA loci associated with the regulation of target gene expression, as well as their potential to modulate the global chromatin structure. In order to do this, transcription factors often interact with other factors and exhibit a certain binding grammar. However, due to both experimental and computational limitations, it is challenging to study the global characteristics of transcription factor binding. Especially at early developmental timepoints, where the input material is sparse, many aspects of transcription factor binding remain obscure. The objective of this thesis is to develop computational methods to study the effect of transcription factor binding during developmental processes. This thesis presents the description of two bioinformatics tools suitable for identifying individual transcription factor binding sites and characterizing the grammar of co-occurring binding events. The first tool, named TOBIAS, is able to identify transcription factor binding using the method of ATAC-seq footprinting. Using this tool, it was possible to create a sequential map of transcription factor activity throughout the early cell divisions of human and mouse embryos. The second tool, TF-COMB, is a method to perform genome-wide transcription factor association analysis, which can utilize binding sites from TOBIAS or other methods. Investigations of both experimental and in silico data across multiple cell types showed that transcription factors bind DNA in the vicinity of other DNA-binding proteins, bind at sites labeled by specific chromatin marks and exhibit preferred binding conformations. In conclusion, these two tools, published in two separate papers, provide a significant improvement to the existing bioinformatics methods for studying global transcription factor binding characteristics in the context of differentiation.