Categorizing the dispersion of open chromatin usage patterns of Treg related genes using an variational autoencoder based algorithm

Abstract

Cellular programs are controlled by the coordinated interactions between transcription factors (TF) and cis-regulatory genomic sequences, including enhancer elements which can be far from the TF they regulate. This enables multiple enhancers to control a single gene, allowing complex gene expression modules. ATAC-seq and single-cell ATAC-seq methods are utilized to assess the gene regulations of cis-regulatory elements represented by open chromatin regions (OCRs), but there is an insufficient exploration of the cis-regulatory architectures in particular genes. In the context of regulatory T cells (Treg), core moderators of immune homeostasis and inflammation, how the transcription factors work together to determine their niches and locations still needs to be determined. This thesis analyzes the OCR activity patterns of individual genes in SVD and UMAP dimensionally reduced datasets retrieved from Splenic T cell scATAC-seq, using a combination of global metrics, OCR clustering algorithms, and variational autoencoders. By classifying the extent of OCR homogenous organization, some genes that interact with Foxp3 and AIRE and those corresponding to activated Treg cells have distinctive OCR activity patterns. This method enables us to picture whether genes are modulated using similar OCR usage patterns regardless of whether there is a direct overlap of the OCRs used and applies to contexts of other immune cells and T cell subtypes.