ChBE McCormick Northwestern
Lab Affiliations: CSB NICO IBiS CLP QBIO

The Bagheri Lab integrates experimental data with computational strategies to elucidate fundamental properties governing intracellular dynamics and intercellular regulation. Our group is highly collaborative and integrates a diverse array of research interests. We take on grand challenges spanning complex dynamics of cell populations, to experimental design and tool development. A common thread that persists among our projects involves elucidating, predicting, and ultimately controlling biological response, particularly in context of disease. When the regulation, or control, of biological function fails, people can manifest a variety of illnesses including cancer and autoimmune disease.

Ongoing collaborative projects span three major thematic areas:

Emergent Phenomena

Multiscale, multiclass models of cell populations

To effectively understand and predict cell population responses to intrinsic perturbations and extrinsic intervention, the integration of intracellular signaling with intercellular dynamics is necessary. Through agent-based models, we simulate and predict the behavior of heterogeneous cell populations. Our in silico framework (developed by Jessica S. Yu) effectively elucidates how changes at the subcellular level emerge into varied population level responses.

Solid tumor microenvironments

Jessica S. Yu

The tumor microenvironment comprises of several moving parts that can both support tumor growth and suppress therapeutic interventions. Resolving how this "whole is greater than the sum of its parts" remains elusive. By integrating a spectrum of tumor cell agents into the context of a heterogeneous healthy agent population, we stand to resolve–at least part–of this mystery. Our agent-based model contains healthy and cancer cell agents, containing metabolism and signaling modules, within a 2D and 3D tissue environment with glucose, oxygen, TGFα, and dynamic vascular function and structure. Interrogation of simulated dynamics can help identify control schemes that have lasting impact on the tumor microenvironment. Our aim is to identify new therapeutic strategies that overcome heterogeneity-derived resistance and sidestep endogenous cellular control schemes.

The impact of tunable features of CAR T-cell therapies in solid tumors

Alexis N. Prybutok

Chimeric antigen receptor (CAR) T-cell therapies have several designable features at multiple levels. Each design choice affects the engineered cells’ ability to target and challenge cancer cells in patients. The challenge: penetrating solid tumors. In collaboration with Prof. Josh Leonard, we aim to identify how changing tunable properties of engineered cells impacts both individual and emergent population-level responses in a solid tumor microenvironment. By using an agent-based model, we can systematically probe tunable, individual and population-level CAR T-cell features to observe emergent, multiscale behavior and determine the optimal set of therapeutic properties required to clear a given tumor.

Predicting drug resistance as a function of tumor heterogeneity

Zeynab Mousavikhamene

Resistance to chemotherapy is a crucial concern in anti-cancer drug treatment, as it causes failure in more than 90% patients with metastatic cancer. Tumor heterogeneity is a major cause of drug resistance. We use agent-based models to predict how and when drugs, with specific mechanisms of action, fail as a function of both individual tumor and cell population heterogeneity within a simulated microenvironment. Results can inform when therapeutic interventions might fail and how to design therapies according to properties of tumor heterogeneity.

Development of integrated synthetic receptors

Kate E. Dray

It is now possible to engineer cells to respond to specific environmental cues in user-defined ways. Although there exist numerous technologies for engineering specific cellular functions, it remains challenging to predict the impact of individual functions, let alone integrate their utility. In collaboration with Prof. Joshua Leonard, we aim to develop a computational framework to facilitate quantitative comparisons across synthetic receptors to enable matching of biosensing goals with receptors that utilize different sensing modalities. Our integrated experiment-modeling characterization will facilitate the development of multi-receptor systems with unprecedented specificity.

Microbe-host interactions

Jason Y. Cain

To develop methods for manipulating the gut microbiome, it is necessary to understand physical properties governing gut colonization of heterogenous microbe populations. Through agent-based models, we can identify properties of cellular metabolism and signaling that support the population-level fitness of microbes. By integrating microbe models with host immune response models, we can design an in silico microenvironment to mimic physiological environments that are otherwise challenging to observe experimentally. Model analysis can be used to identify key properties that influence colonization potential.

Circadian regulation

Narasimhan Balakrishnan

Models for circadian rhythms have been well explored in the context of single cell behavior. How do populations of heterogeneous clock neurons behave when coupled? We investigate the emergent functional consequences of cellular and structural heterogeneity on networks of coupled circadian oscillators. Identifying how cellular heterogeneity impacts circadian performance and phase entrainment can inform control strategies that correct circadian misalignment and provide guidelines to better design synthetic biological oscillator systems.

Dynamical Systems

Innate immune signaling

Joseph J. Muldoon

During injury or infection, the innate immune system initiates a response that is regulated tightly to counter the perceived threat while also limiting toxicity to the host. Seemingly counterintuitive to this organism-level objective, we observe that cells that are genetically identical can respond heterogeneously to the same cue. In collaboration with Prof. Josh Leonard, we investigate these differences and their functional consequences, by combining multiparametric single-cell time-course analyses, mechanistic nonlinear differential equation modeling, and genome-scale experiments.

Optimizing ex vivo megakaryocyte differentiation

Jia J. Wu

Cell fate commitment is orchestrated by complex molecular signals modulated by transcription factors (TFs). Insights into TF network dynamics that govern maturation can reveal opportunities to drive expression toward specific lineages. In particular, histone deacetylase sirtuin-1 (SIRT1) has been shown to potently alter the phenotype and cell lineage fate via chromatin remodeling, although SIRT1’s role in mediating hematopoietic transcription factors remains unclear. In collaboration with Prof. Bill Miller, we investigate the role of SIRT1 deacetylation on the sensitivity of TF-TF interactions that drive megakaryocytic differentiation. To identify causal relationships and sensitivities of these interactions, we employ systematic perturbations and profile differentially expressed TFs over the course of differentiation.

Redundancy in gene regulatory networks

Sebastian M. Bernasek

During the development of complex organisms, gene regulatory networks integrate external signals to ensure the timely execution of downstream events. These networks exhibit extensive functional redundancy, enabling cells to partially compensate for deleterious mutations. We seek to elucidate evolutionary forces shaping these topologies by identifying additional benefits that redundant regulatory mechanisms confer upon populations of organisms. In collaboration with Prof. Luis Amaral and Prof. Rich Carthew, our approach couples model-based predictions with experimental validation in model organisms in an effort to generate new hypotheses.

Designing synthetic genetic programs

Joseph J. Muldoon

Advances in genetic engineering are enabling the development of mammalian cell-based devices for therapeutic and diagnostic applications. As increasingly sophisticated functions are sought, there remains a fundamental need to make cells carry out user-defined instructions precisely and safely. In collaboration with Prof. Josh Leonard, we develop dynamical and statistical models to characterize the properties of newly developed molecular components such as synthetic receptors and transcription factors, and predict how they can be integrated effectively to produce customizable sense-and-respond behaviors.

Cell differentiation during eye development

Sebastian M. Bernasek

Transcription factors coordinate the timing and execution of cell differentiation by tuning the expression of target genes. Some transcription factors promote differentiation, while others impede it. Competing transcription factors are often co-expressed in vivo, and it remains unclear how cells reliably integrate their antagonistic inputs. We combine confocal microscopy with computer vision techniques to measure the expression dynamics of competing transcription factors in the developing fruit fly eye. Our approach facilitates quantitative modeling and characterization of transcription factor activity both before, during, and after differentiation.

Decision making in Drosophila navigation

Josh I. Levy

Animals use sensory information from the environment to avoid harm and ensure survival. Our work seeks to uncover the essential neural circuitry for processing of simple environmental stimuli and execution of behaviors. In this collaboration with Prof. Marco Gallio, we couple experimental analysis with methods from manifold learning and mathematical modeling to interrogate the nature of animal navigation and its fundamental connection to underlying circuitry.

Network Inference, Machine Learning, & Tool Development

Network inference performance

Joseph J. Muldoon | Jessica S. Yu

Network inference algorithms aim to uncover regulatory interactions underlying cell decision-making, disease progression, and therapeutic interventions. These tools have proven invaluable in studying differentiation, identifying regulators and their targets in diseases such as cancer, and predicting drug response mechanisms. Having an accurate blueprint of regulation is important for understanding and controlling cell behavior. We evaluate the performance of network inference algorithms as a function of topology, logic, kinetic parameters, and sampling parameters.

Cell signaling dynamics in cancer

Justin D. Finkle

Cells integrate multiple signals to effectively respond to their environments. In collaboration with Prof. Jonathan Licht, we investigate the dynamic role of Sprouty in cell signaling and cancer. Sprouty is a feedback regulator of receptor tyrosine kinase (RTK) pathways. The dynamics of Sprouty expression encode its regulatory function and mechanism. Using new statistical, network inference, and visualization methods, we identify oncogenic targets of Sprouty that might confer Sprouty's tumor suppressing activity.

Mapping the stress response from genome to phenome

Jason Y. Cain

Environmental stressors trigger multiscale responses spanning subcellular networks to organismal physiology. Analyzing transcriptomics, proteomics, and phenotypic response data enables us to identify associations between transcription factors and proteins as well as factors or regulatory motifs that explain organism-level responses. We collaborate with Profs Anne Todgham and Lars Tomanek to identify how subceullar networks and organism-level tissue phenotypes regulate stress response. By employing statistics and network inference, we identify links in multiscale stress responses in Mytilus californianus.

Feature extraction in flow cytometry data

Jia J. Wu

Manual analysis of temporal trends is both labor intensive and prone to bias. We have developed a new computational algorithm to quantify and predict temporal trends of developing cell subpopulations in flow cytometry data. We employ our new tool to identify influential factors for multi-phase hematopoietic-stem-cell-to-megakaryocyte differentiation. In particular, production of target cell types is highly dependent on donor characteristics of starting input cells. Our method extracts predictive features from Gaussian mixtures to generate models for time-course flow cytometry data.

Magnetic barcode imaging

Slava S. Butkovich

Accurate and precise quantification of chemical species can greatly improve patient outcomes. Magnetic barcode imaging (MBI), a recently developed version of magnetic resonance imaging (MRI), is a promising machine-learning-dependent and multiplexed approach for quantifying chemical species. In collaboration with Prof. Evan Scott, we aim to develop pipelines for MBI that use various tissue backgrounds to identify potential disease states. Demonstrating that MBI can be applied to diverse tissue types would greatly expand its clinical capabilities, potentially allowing for pretreatment of diseases before traditional symptoms are observed.

Cytoskeletal dynamics as a cancer diagnostic

Zeynab Mousavikhamene

A multitude of cytoskeletal organization and surface adhesive characteristics are prevalent across different cancer cell lines. In collaboration with Prof. Milan Mrkisch, we quantify these characteristics as a function of micro-patterning to reduce cell heterogeneity and control cell shape. We aim to develop supervised machine learning models that can identify features of cytoskeletal structure that discriminate among cancer cell types and stages of metastasis.

Predicting structure-activity relationships

Albert Y. Xue

Improved characterization and prediction of structure-activity relationships would benefit our understanding of human diseases and biology, enable greater control of engineered therapeutics, and help diagnose diseases with greater accuracy. However, enzyme characterization has until recently remained limited due to a lack of high-throughput methods to gather comprehensive data. In collaboration with Prof. Milan Mrksich, we employ SAMDI mass spectrometry data to quantify peptide-enzyme activities on high throughput peptide arrays. We apply machine learning and statistical tools to extract enzyme binding patterns and create models that predict peptide-enzyme binding activities.