Mechanistic modeling and systems analysis across biological scales, from single cells and regulatory networks to microbial communities

Mahajan, Tarun

Permalink

https://hdl.handle.net/2142/127154 Copy

Description

Title

Mechanistic modeling and systems analysis across biological scales, from single cells and regulatory networks to microbial communities

Author(s)

Mahajan, Tarun

Issue Date

2024-09-30

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

Maslov, Sergei

Doctoral Committee Chair(s)

Maslov, Sergei

Committee Member(s)

• Irudayaraj, Joseph M.K.
• Zhao, Sihai D
• Dar, Roy D

Department of Study

Bioengineering

Discipline

Bioengineering

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Gene regulatory network
• biophysical modeling
• transcriptional bursting
• single cell RNA sequencing
• RNA velocity
• microbial community
• gut microbiome

Abstract

Mechanistic models are essential in biology, enabling predictions across scales from single cells and gene regulatory networks to complex microbial communities. These models are vital for applications in drug discovery, disease understanding, gut microbiome health, and ecosystem restoration. The appropriate model choice depends on the scale and nature of the biological processes; for instance, stochastic models suit discrete processes like mRNA transcription, while ODE-based models are better for complex microbial communities. This dissertation investigates mechanistic and regulatory network models across three biological scales: single cells, gene regulatory networks, and microbial communities. We first examined a model of bursty gene expression in single cells, demonstrating that the longer half-life of eukaryotic proteins compared to mRNA prevents inference of transcriptional regulatory networks (TRN) from steady-state single-cell RNA data. We validated this prediction with real yeast single-cell RNA sequencing (scRNA-seq) data. Next, we utilized the biophysical model of bursty gene expression to develop a method called noSpliceVelo, which infers RNA velocity from scRNA-seq data without relying on splicing dynamics. noSpliceVelo leverages the temporal relationship between the variance and the mean of gene expression to infer RNA velocity, operating on the time scale of mRNA degradation, which is comparable to developmental processes. As a result, noSpliceVelo avoids issues associated with splicing-based RNA velocity techniques, such as biased and inaccurate intronic reads, a shorter time scale of splicing relative to developmental processes, and the absence of transcriptional bursting in the model. Our method accurately predicted the temporal evolution of transcriptional states in various cellular lineages in real scRNA-seq data and inferred biophysically relevant parameters for transcriptional regulation. noSpliceVelo either matches or outperforms commonly used splicing-based RNA velocity methods. At the scale of regulatory networks, we investigated the joint organization of TRN and protein-protein interaction (PPI) networks across multiple species. We showed that TRNs and PPIs are non-randomly coupled in five different eukaryotic species, and this coupling enhances the robustness of the multiplex network to targeted attacks. Furthermore, we demonstrated that functionally important genes and proteins, such as those essential, disease-related, or interacting with pathogen proteins, are preferentially located in critical parts of the human multiplex. Finally, at the largest scale, we developed and analyzed a simplified consumer-resource model for complex microbial communities in serial dilution experiments. We fitted the model to describe serial dilution experiments in a defined synthetic human gut microbiome containing over 60 strains growing in several hundred resources. Our model accurately predicted the serial dilution dynamics of the community, with a correlation coefficient between predicted and observed strain abundances as high as 0.8. The model also enabled us to perform three different in silico perturbation experiments to: (i) study the interaction network between the strains; (ii) explore direct and indirect interactions between strains and resources; and (iii) develop a resource supplementation protocol to maximally equalize steady-state strain abundances.

Graduation Semester

2024-12

Type of Resource

Thesis

Handle URL

https://hdl.handle.net/2142/127154

Copyright and License Information

Copyright 2024 Tarun Mahajan

Owning Collections