Integrative analysis of transcriptional regulatory programs in mammalian systems

Tabe Bordbar, Shayan

Permalink

https://hdl.handle.net/2142/113031 Copy

Description

Title

Integrative analysis of transcriptional regulatory programs in mammalian systems

Author(s)

Tabe Bordbar, Shayan

Issue Date

2021-07-13

Director of Research (if dissertation) or Advisor (if thesis)

Sinha, Saurabh

Doctoral Committee Chair(s)

Sinha, Saurabh

Committee Member(s)

• Peng, Jian
• Vasudevan, Shobha
• Halfon, Marc

Department of Study

Computer Science

Discipline

Computer Science

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

Gene Regulatory Network, non-coding RNA, Gene Expression Prediction

Abstract

Gene regulatory network (GRN) is the most central known information processing program of life. To understand its importance and complexity, it is sufficient to note that all the cells in our body contain the same genetic code yet carry out functions ranging from energy storage to immune response. Aberrations in GRNs are known to be implicated in multitude of biological disorders, yet we are far from a complete understanding of the involved players and the mechanisms through which they function. In order to study the workings of GRNs, we need to identify, and quantify the input and output of this information processing system and make sense of the relationship between input and output. Inputs to the system are first the sequence of DNA, and second the collection of DNA-interaction maps for macro-molecules that come into contact with DNA. The immediate biosynthetic outcome of the system can be measured as the transcriptomic profile of the cell (i.e., count of RNA molecules transcribed from each DNA region of some particular length). The goal of this thesis is to provide clues to understand how the cell processes information from external stimuli in light of its genetic background, leading to a harmonic transcriptional response. My PhD work comprises of three studies focused on understanding of gene regulatory networks and evaluation of their plausibility. Meaningful regulatory relationships in GRNs are expected to generalize to distinct cellular conditions. We address the challenge of generalizability evaluation in gene expression prediction task through quantifying the distinctness of test and training sets in this context. We introduce a novel cross-validation strategy that takes generalizability into account. In the next chapters, I focused on mechanistic understanding of two known players in GRNs, namely Transcription Factors (TFs) and long non-coding RNAs (lncRNAs). More specifically, taking on the challenge of sequence to expression prediction in mammalian systems, we report the first thermodynamics-based model of transcription in breast-cancer cells. Through combinatorial modeling of TF roles in observed transcriptional aberrations of breast cancer cells, we identified literature-supported roles for many contributing TFs. Such models enabled us to make mechanism-aware variant impact predictions, several of which were then experimentally verified. Finally, we study the mechanisms underlying the interactions between lncRNAs and chromatin. Given the experimental loss of function studies and considering the RNA-first hypothesis, it is hard not to imagine a central role for RNAs in fundamental cellular processes such as transcription and DNA repair. However, poor sequence conservation of lncRNAs has prevented researchers from identifying core elements involved in their function. In the last chapter of this thesis we explore the sequence and context landscape of mouse Embryonic Stem Cells (mESC) to identify factors predictive of lncRNA-chromatin interactions. Through a classification-based framework we evaluate the predictive value of various families of features and identify clues linking lncRNA-chromatin interactions with DNA methylation, DNA Damage Response and transcriptional machinery. In conclusion, through this work we identify factors involved in RNA-DNA interactions, introduce a mechanistic framework to model transcriptional regulation by TFs, as well as a novel method to assess the generalizability of such models.

Graduation Semester

2021-08

Type of Resource

Thesis

Permalink

http://hdl.handle.net/2142/113031

Copyright and License Information

2021 Shayan Tabe Bordbar

Owning Collections