Organized by: Kang-Ling Liao (University of Manitoba), Wenrui Hao, Pennsylvania State University

Note: this minisymposia has multiple sessions. The other session is: and Part-2.

1. Weitao Chen University of California, Riverside
   "A Mechanochemical Coupled Model to Understand Budding Behavior in Aging Yeast"
Cell polarization, in which a uniform distribution of substances becomes asymmetric due to internal or external stimuli, is a fundamental process underlying cell mobility and cell division. Budding yeast provides a good system to study how biochemical signals and mechanical properties coordinate with each other to achieve stable cell polarization and give rise to certain morphological change in a single cell. Recent experimental data suggests yeast budding develops into two trajectories with different bud shapes as mother cells become old. We first developed a 2D model to simulate biochemical signals on a shape-changing cell and investigated strategies for robust yeast mating. Then we extended and coupled this biochemical signaling model with a 3D subcellular element model to take into account cell mechanics, which was applied to investigate how the interaction between biochemical signals and mechanical properties affects the cell polarization and budding initiation. This 3D mechanochemical model was also applied to predict mechanisms underlying different bud shape formation due to cellular aging.
2. Harsh Jain University of Minnesota Duluth
   "Looking Beyond Data: Simulating Treatment Outcomes for Unobserved Heterogeneous Populations Using Preclinical Insights"
Developing new cancer drugs involves significant investments of time and resources, yet many promising candidates fail during clinical trials. One potential reason for this failure is that preclinical testing typically relies on genetically identical animals and uniform cell lines, which do not reflect the diversity found in actual patient populations. Additionally, preclinical data is often presented in aggregated form, masking important individual-level differences that could inform clinical predictions. In this talk, I will present a case study of non-small cell lung cancer xenograft treatment with radiation to introduce our Standing Variations Modeling approach, which addresses these issues in two main steps. First, we deconstruct aggregated preclinical data – specifically, average tumor volume time-courses and Kaplan-Meier survival curves – to recover individual-level variation, uncovering hidden differences among study subjects (“who’s in”). Second, we use these insights to simulate treatment outcomes for broader, more diverse virtual populations through computational modeling (“who’s out”). A key innovation in our method is the assignment of a personalized survival probability to each virtual participant, explicitly linked to their unique disease dynamics. This mechanistic connection allows us to capture inter-individual variability and supports meaningful extrapolation to unobserved populations. By moving beyond aggregate data and homogeneous preclinical models, this approach offers a more nuanced and practical path to clinical translation.
3. Leili Shahriyari University of Massachusetts Amherst
   "Data Driven QSP Modeling of Cancer: A Step Toward Personalized Treatment"
Our work explores the possibility of creating personalized mathematical models for cancer to better understand the progression of an individual's cancer. By simulating the unique characteristics of each tumor and its response to treatments, we aim to offer insights into personalized cancer care. Our method combines elements of mechanistic modeling and machine learning techniques to create individualized predictions. A central aspect of our approach is the use of a mechanistic model based on quantitative systems pharmacology (QSP). QSP is a computational method used to analyze drug interactions and effects, and it plays a crucial role in our project. The model includes a large system of nonlinear equations modeling both bio-chemical and bio-mechanical integrations in the tumors. We acknowledge that a common challenge in QSP modeling is accurately determining parameters, especially since traditional models often assume a general uniformity across different patients' diseases. This assumption can lead to limitations when calibrating parameters using varied data sources. Our objective is to build a more personalized mathematical framework by concentrating on individual patient data for parameter estimation. We adjust the QSP model parameters for each patient based on their unique data. Through detailed sensitivity analysis and uncertainty quantification, we identify key interactions in the model and define the range of our predictions. By integrating this tailored QSP model with insights into cellular and molecular interactions, we hope to better predict how cancer evolves and responds to specific treatments. We are excited about the potential this has for advancing personalized cancer therapy, though we are aware of the challenges and complexities involved in this endeavor.
4. Nourridine Siewe Rochester Institute of Technology
   "Osteoporosis Induced by Cellular Senescence: A Mathematical model"
Osteoporosis is a disease characterized by loss of bone mass, where bones become fragile and more likely to fracture. Bone density begins to decrease at age 50, and a state of osteoporosis is defined by loss of more than 25%. Cellular senescence is a permanent arrest of normal cell cycle, while maintaining cell viability. The number of senescent cells increase with age. Since osteoporosis is an aging disease, it is natural to consider the question to what extend senescent cells induce bone density loss and osteoporosis. In this paper we use a mathematical model to address this question. We determine the percent of bone loss for men and women during age 50 to 100 years, and the results depend on the rate η of proliferation of senescent cell, with η=1 being the average rate. In the case η=1, the model simulations are in agreement with empirical data. We also consider senolytic drugs, like fisetin and quercetin, that selectively eliminate senescent cells, and assess their efficacy in terms of reducing bone loss. For example, at η=1, with estrogen hormonal therapy and early treatment with fisetin, bone density loss for women by age 75 is 23.4% (below osteoporosis), while with no treatment with fisetin it is 25.8% (osteoporosis); without even a treatment with estrogen hormonal therapy, bone loss of 25.3% occurs already at age 65.

Organized by: Sebastian Schreiber (University of California, Davis)

Note: this minisymposia has multiple sessions. The other session is: and Part-2.

1. Alanna Hoyer-Leitzel Mount Holyoke College
   "Resilience to reinfection in an impulsive model of viral exposure"
Re-exposure to virus in an ongoing, low-level endemic can lead to the appearance of long term immunity in an individual. Starting with an ordinary differential equations model for an immune system, we simulate repeated viral re-exposure with a discrete impulse of virus. When the re-exposures are deterministic, we can find different long term outcomes of either reinfection or protection, depending on viral dose size and frequency. We investigate how these outcomes persist when the re-exposures are stochastic. We examine the effects of the choice of probability distributions for viral dose size and frequency.
2. Jakob Kaare-Rasmussen University of California, Davis
   "Habitat Destruction and Disturbance in Forest Ecosystems"
Forests around the world are increasingly threatened by habitat destruction and disturbances—factors that, together, can have profound and often unexpected effects on these ecosystems. Habitat destruction is the loss of habitat due to urbanization or agricultural lands while disturbances are perturbations of the system that leave the environment habitable, like forest fires and drought. Forests are not just collections of trees; they also depend on below-ground mycorrhizal fungal mutualists. The mycorrhizal fungi facilitate the uptake of nutrients and water for the trees while receiving products of photosynthesis in return. Despite their importance and close relationship with trees, mycorrhizal fungi are often overlooked in mathematical models of forest response to environmental stressors. To address this gap, I developed a metacommunity model that explicitly incorporates the mutualism between trees and mycorrhizal fungi. After analyzing the dynamics of the unperturbed system, I examined the impact of habitat destruction—modeled as the continuous loss of habitat over time—which can lead to catastrophic forest collapse. This habitat destruction is often accompanied by disturbance events, modeled here as discrete events that “kick” the system’s state. The interplay between continuous background change (destruction) and sudden disturbances (kicks) generates complex and sometimes counterintuitive behaviors, including rate-induced tipping from a healthy forested state to local extinction. To gain insights into the dynamical mechanisms underlying this rate-tipping, I analyzed a simplified one-dimensional bistable model. This reduced model reveals general patterns relevant to a wide range of systems experiencing both gradual environmental change and discrete disturbance events. Given that many ecological systems are facing similar pressures, such as ongoing environmental degradation combined with frequent disturbances, this work illustrates how non-autonomous flow-kick models can be used to better understand and predict how ecosystems respond to these dual stressors.
3. Vanja Dukic University of Colorado
   "Weak-form inference for hybrid dynamical systems in ecology"
Species subject to predation and environmental threats commonly exhibit variable periods of population boom and bust over long timescales. Understanding and predicting such behaviour, especially given the inherent heterogeneity and stochasticity of exogenous driving factors over short timescales, is an ongoing challenge. A modelling paradigm gaining popularity in the ecological sciences for such multi-scale effects is to couple short-term continuous dynamics to long-term discrete updates. We develop a data-driven method utilizing weak-form equation learning to extract such hybrid governing equations for population dynamics and to estimate the requisite parameters using sparse intermittent measurements of the discrete and continuous variables. The method produces a set of short-term continuous dynamical system equations parametrized by long-term variables, and long-term discrete equations parametrized by short-term variables, allowing direct assessment of interdependencies between the two timescales. We demonstrate the utility of the method for epizootics experienced by the North American spongy moth (Lymantria dispar dispar). Joint work with Dan Messenger and Greg Dwyer.
4. Punit Gandhi Virginia Commonwealth University
   "The impact of rainfall variability on pattern formation in a flow-kick model for dryland vegetation bands"
Water input in dryland ecosystems comes in the form of infrequent, discrete and largely unpredictable rainstorms. These short-lived pulses are known to sustain large-scale spatial patterns that appear as regularly spaced bands of dense vegetation separated by regions of bare soil. I will present a flow-kick modeling framework for such dryland vegetation patterns that treats storms as instantaneous kicks to the soil water, which then interacts with vegetation during the long dry periods between the storms. The spatial profiles of the nonlocal, state-dependent soil water kicks capture positive feedbacks in the storm-level hydrology that act to concentrate water within the vegetation bands. This flow-kick model predicts that variance in rainfall, introduced through randomness in the timing and magnitude of water input from storms, decreases the parameter range over which patterns form and may negatively impact ecosystem resilience. Authors: Matthew Oline (University of Chicago), Mary Silber (University of Chicago)

Organized by: Cody FitzGerald (Northwestern University), Rainey Lyons (CU Boulder), Nora Heitzman-Breen (CU Boulder), Susan Rogowski (NCSU)

Note: this minisymposia has multiple sessions. The other sessions are: Part-2, Part-3, and Part-4.

Note: this minisymposia has been accepted, but the abstracts have not yet been finalized.

Organized by: James Osborne (University of Melbourne), James Glazier (Indiana University) Yi Jiang (Georgia State University)

Note: this minisymposia has multiple sessions. The other sessions are: Part-2, and Part-3.

1. Yi Jiang Georgia State University, USA
   "Multicellular Modelling of Collective Cancer Invasion"
Collective invasion enhances cancer metastasis. However, the mechanisms underlying cancer collective invasion remains poorly understood. I will discuss two types of collective invasion, one with leaders and followers that engage in mutualistic social interactions, the other involves hypoxia induced secretome. We use cell-based multiscale models to elucidate the physical mechanisms for the emergence of collective behavior. In the leader-follower invasion, cell-cell adhesion and cell migration are the key drivers of migration patterns, while in the microenvironment-induced invasion, protrusion force and adhesion together give rise to symmetry breaking and directional migration. The results from the experimental and computational models combined provide new insights into tumor metastasis in terms of tumor heterogeneity and cellular response to microenvironmental stimuli
2. Jupiter Algorta University of British Columbia, CANADA
   "Simulating Cell Decisions and Embryo Structure with Morpheus"
One of the most fascinating aspects of early embryonic development is how a single cell gives rise to a structured, multicellular organism. In mammals, this process begins with a few identical cells that divide and gradually acquire distinct identities. These identities not only depend on each cell’s own internal machinery but also emerge from how cells interact with their neighbors and their environment. In this project, we modeled two critical stages of this developmental process. First, cells differentiate between becoming part of the outer layer (which contributes to the placenta) or the inner group (which forms the embryo proper). This initial decision is governed by how cells contact one another. Second, cells in the inner group further specialize, influenced by a signaling molecule called FGF4, which spreads through the surrounding space and nudges cells toward one of two fates: forming the future embryo (epiblast) or a supporting layer (primitive endoderm). To capture this complex cascade of decisions, we translated two detailed models by De Mot et al. (2016) and Cang et al. (2021) into a unified spatial simulation using Morpheus, a platform for multiscale modeling. These existing models describe how genes interact within each cell through systems of equations, while our goal was to bring them into a spatial context, allowing cells to move, divide, and interact in space, while still carrying out their internal decision-making logic. While the original models include dozens of interacting components and parameters, Morpheus’ design made it possible to integrate these internal processes with the physical layout and behaviour of the cells. Each simulated cell runs its own internal 'program' while also communicating with others through contact or diffusing signals in the surrounding space. The platform’s graphical interface and modular setup made this a manageable task, even for undergraduate researchers. Our resulting simulation recreates known patterns of cell arrangement and fate specification seen in real embryos. More importantly, this case study illustrates how modeling tools like Morpheus can help translate complex biological mechanisms into testable, visual models, even when starting from dense theoretical descriptions.
3. Andreas Buttenschoen University of Massachusetts, USA
   "Robust Numerical Methods for cells invading extracellular matrix: Adaptive Time-stepping and preconditioning for reproducible multicellular models"
Cell migration through extracellular matrix (ECM) environments represents a fundamental biological process essential for development, immune response, wound healing, and cancer metastasis. This migration presents significant physical challenges as cells must simultaneously use the ECM as a substrate for force transmission while overcoming its role as a mechanical barrier. In this talk, I will present a physics-based computational model that elucidates how cells employ three primary 'space negotiation' strategies to navigate dense ECM: (1) adaptive cellular deformation, (2) mechanical remodeling of surrounding matrix, and (3) enzymatic degradation via matrix metalloproteinases (MMPs). Our model captures the essential mechanical interactions between deformable rod-shaped cells and a viscoelastic fiber network, with cells extending filopodia that establish adhesion sites and generate traction forces. Through systematic computational analysis, we demonstrate that cellular migration efficiency exhibits a biphasic response to ECM density, with optimal migration occurring at intermediate pore sizes that match nuclear dimensions. We further show that different microenvironmental contexts necessitate distinct combinations of space negotiation strategies - while ECM degradation is dispensable in pre-formed tracks, it becomes essential in dense matrices where nuclear size represents the primary migration-limiting factor. In the final portion of this talk, I will discuss the numerical methods that enable robust and reproducible simulation of these complex multicellular systems. Specifically, I will present our implementation of adaptive time-stepping using embedded Runge-Kutta methods that allow users to specify absolute and relative error tolerances, ensuring reliable integration of agent-based models. Additionally, I will describe graph-based preconditioning techniques for efficiently solving the overdamped Langevin dynamics with anisotropic friction tensors, and discuss memory management strategies using smart pointers and double buffering that significantly improve computational performance. These numerical advances enable simulation of larger multicellular collectives while maintaining mechanistic fidelity at the single-cell level.
4. Rajendra Singh Negi Syracuse University, USA
   "Multicellular modeling of how myosin localization impacts symmetry-breaking in zebrafish embryonic development"
A fundamental question is how organisms control cell and organ morphology during development, and we address this question using Kupffer’s Vesicle (KV), the left-right organizer in zebrafish, as a simple model organ. Both the cells that comprise the KV, and the organ itself, change shape in a stereotyped manner that is important for organ function. While multiple mechanisms have been proposed to govern these shape changes, recent studies combining 3D simulations with laser ablation experiments have shown that the slow movement of KV through the surrounding tailbud tissue generates dynamic forces that alter the organ and cell shape [1]. To understand the molecular mechanisms that affect this motion, we have developed an experimental protocol to perturb myosin activity in a localized region of interest in the tailbud, using an optically controlled rho-kinase inhibitor. We implement a 3D vertex-based simulation framework that captures the multicellular dynamics of KV migration. Our model incorporates key mechanical interactions: posterior traction from migratory cells, anterior pushing by the notochord using conversion-extension, and viscoelastic responses from surrounding tailbud tissue. We model the effect of the caged rho-kinase inhibitor as a diffusing patch of signal that emanates from a region of interest below the tailbud, which alters both the dynamic forces applied to KV as well as the mechanics of the tailbud tissue. This approach allows us to investigate how localized molecular perturbations propagate through tissue to influence organ motion and morphology. Preliminary results show that such perturbations can alter the motion and shape of KV, revealing how spatially confined molecular changes can drive large-scale morphogenetic transformations. [1] Manna et al. bioRxiv, https://arxiv.org/pdf/2407.07055 This work was supported by NIH R01HD099031.

Organized by: Sebastian Schreiber (University of California, Davis)

Note: this minisymposia has multiple sessions. The other session is: and Part-1.

1. Junping Shi College of William and Mary
   "Effect of rotational grazing on plant and animal production"
It is a common understanding that rotational cattle grazing provides a better yield than continuous grazing, but a qualitative analysis is lacking in the agriculture literature. In rotational grazing, cattle periodically move from one paddock to another in contrast to continuous grazing, in which the cattle graze on a single plot for the entire grazing season. Here we quantitatively show how production yields and stockpiled forage are greater in rotational grazing in some harvesting models. We construct a vegetation grazing model on a fixed area, and by using parameters obtained from agricultural publications and keeping the minimum value of remaining forage constant, our result shows that both the number of cattle per acre and stockpiled forage increase for all tested rotational configurations than the continuous grazing. Some related spatial harvesting models are also discussed. This is a joint work with Mayee Chen.
2. Kate Meyers Carleton College
   "From deluges to drizzle: continuous limits of flow-kick models"
To incorporate ongoing disturbances into a differential equation (DE) model of biological processes, one might embed the disturbance continuously in the DE or resolve the disturbance discretely. In this talk we’ll explore the flow-kick approach to modeling repeated, discrete disturbances and examine the dynamic implications of this modeling choice. We’ll position continuous disturbances as limits of repeated, discrete ones and share recent results on how flow-kick systems both mimic and depart from their continuous analogs.
3. Rebecca Tyson University of British Columbia, Okanagan
   "Host-parasitoid systems are vulnerable to extinction via P-tipping: Forest Tent Caterpillar as an example"
Continuous-time predator-prey models admit limit cycle solutions that are vulnerable to the phenomenon of phase-sensitive tipping (P-tipping): The predator-prey system can tip to extinction following a rapid change in a key model parameter, even if the limit cycle remains a stable attractor. In this paper, we investigate the existence of P-tipping in an analogous discrete-time system: a host-parasitoid system, using the economically damaging forest tent caterpillar as our motivating example. We take the intrinsic growth rate of the consumer as our key parameter, allowing it to vary with environmental conditions in ways consistent with the predictions of global warming. We find that the discrete-time system does admit P-tipping, and that the discrete-time P-tipping phenomenon shares characteristics with the continuous-time one: Both require an Allee effect on the resource population, occur in small subsets of the phase plane, and exhibit stochastic resonance as a function of the autocorrelation in the environmental variability. In contrast, the discrete-time P-tipping phenomenon occurs when the environmental conditions switch from low to high productivity, can occur even if the magnitude of the switch is relatively small, and can occur from multiple disjoint regions in the phase plane. This is joint work with Bryce F. Dyck.
4. Sebastian Schreiber University of California, Davis
   "Coexistence and extinction in flow kick systems via Lyapunov exponents"
Natural populations experience a complex interplay of continuous and discrete processes: continuous growth and interactions are punctuated by discrete reproduction events, dispersal, and external disturbances. These dynamics can be modeled by impulsive or flow-kick systems, where continuous flows alternate with instantaneous discrete changes. To study species persistence in these systems, an invasion growth rate theory is developed for flow-kick models with state-dependent timing of kicks and auxiliary variables that can represent stage structure, trait evolution, and environmental forcing. The invasion growth rates correspond to Lyapunov exponents that characterize the average per-capita growth of species when rare. Two theorems are proven that use invasion growth rates to characterize permanence, a form of robust coexistence where populations remain bounded away from extinction. The first theorem uses Morse decompositions of the extinction set and requires that there exists a species with a positive invasion growth rate for every invariant measure supported on a component of the Morse decomposition. The second theorem uses invasion growth rates to define invasion graphs whose vertices correspond to communities and directed edges to potential invasions. Provided the invasion graph is acyclic, permanence and extinction are fully characterized by the signs of the invasion growth rates. Invasion growth rates are also used to identify the existence of extinction-bound trajectories and attractors that lie on the extinction set. To demonstrate the framework's utility, these results are applied to a microbial serial transfer model where state-dependent timing enables coexistence through a storage effect, a spatially structured consumer-resource model showing intermediate reproductive delays can maximize persistence, and an empirically parameterized Lotka-Volterra model demonstrating how disturbance can lead to extinction by disrupting facilitation. Mathematical challenges, particularly for systems with cyclic invasion graphs, and promising biological applications are discussed. These results reveal how the interplay between continuous and discrete dynamics creates ecological outcomes not found in purely continuous or discrete systems, providing a foundation for predicting population persistence and species coexistence in natural communities subject to gradual and sudden changes

Organized by: Cody FitzGerald (Northwestern University), Rainey Lyons (CU Boulder), Nora Heitzman-Breen (CU Boulder), Susan Rogowski (NCSU)

Note: this minisymposia has multiple sessions. The other sessions are: Part-1, Part-3, and Part-4.

Note: this minisymposia has been accepted, but the abstracts have not yet been finalized.

Organized by: James Osborne (University of Melbourne), James Glazier (Indiana University) Yi Jiang (Georgia State University)

Note: this minisymposia has multiple sessions. The other sessions are: Part-1, and Part-3.

1. Claire Miller Auckland Bioengineering Institute, NEW ZEALAND
   "Multicellular modelling of endometrial cell invasion in endometriosis lesion onset"
Endometriosis is a chronic gynaecological condition that is estimated to affect 1 in 9 people with a uterus. The disease is characterised by the presence of cells similar to those that line the uterus (endometrial cells) growing as lesions outside the uterus, such as in the lining of the pelvis. It is hypothesised that the disease originates from menstrual debris entering the pelvic region via the fallopian tubes. The endometrial cells in this menstrual debris then breach the epithelial layer lining the pelvis and form lesions that intrude into the lower layers of the tissue. Very little is understood about the conditions required for endometriosis onset. The endometrial cell invasion behaviour has been hypothesised to be a result of dysfunctions in the immune system, the invading endometrial cells, the breached epithelial layer, or any combination of these. In this talk I will present a multicellular agent-based model for endometrial cell invasion of an epithelial monolayer. Using this model, I will explore several of the hypotheses around disease onset, such as those related to cell proliferation and adhesion, and assess the level to which they promote endometrial cell invasion.
2. TJ Sego University of Florida, USA
   "Quantitative Reproducibility at Scale: A Federated, Standardizable Approach"
Stochastic simulations are commonly used to quantitatively or semi-quantitatively describe the dynamics of biological systems. Proving reproducibility of simulation results is critical to establishing the credibility of a model. However, reproducibility of stochastic simulation is difficult for numerous reasons. For example, under-sampling produces insufficient information to allow conclusive findings from independent reproducibility studies. Hence, along with measures to compare results, reproducible stochastic simulation as a community-level practice requires measures of when results can be verified as meaningfully reproduced in independent study, and data formats for facilitating information exchange. This session presents the Empirical Characteristic Function Equality Convergence Test (EFECT), a data-driven method to quantify the reproducibility of stochastic simulation results. EFECT provides a quantitative measure of the reproducibility of stochastic results, called the EFECT Error, for modelers to determine a sample size that allows independent reproducibility studies. EFECT also provides a statistical test for performing reproducibility studies with an a priori significance, enabling modeling communities to develop standards and best practices. To this end, EFECT defines the minimum necessary information, called an EFECT Report, to facilitate exchange between modelers for reproducibility studies. The session surveys numerous applications that demonstrate EFECT enabling reproducible stochastic simulation with a variety of modeling methodologies, including ordinary differential equations with stochastic parameter sampling, stochastic differential equations, agent-based models, and uncertainty quantification in physics-informed neural networks. The session concludes with a detailed look at how a Python library implementation of EFECT, called libSSR, is enabling quantitative reproducibility in multicellular modeling as a federated, community-level activity.
3. Steve Runser ETH Zurich, SWITZERLAND
   "PolyHoop & SimuCell3D: Efficient and Versatile Tissue Simulations in 2D and 3D"
Accurate simulation of epithelial tissue dynamics requires models that capture complex, polarized cell shapes with high spatial resolution. Previous approaches were hampered by high computational cost or lacked essential biological detail. To overcome these challenges, we developed two powerful new computational frameworks for simulating epithelial tissues in 2D and 3D. PolyHoop [Vetter, Runser & Iber, Comput. Phys. Commun. 299, 109128 (2024)] models cell membranes as closed flexible hoops in 2D, incorporating intra- and intercellular forces and topological events such as cell division and fusion. SimuCell3D [Runser, Vetter & Iber, Nat. Comput. Sci.] extends this approach to 3D, using triangulated surfaces to represent membranes, nuclei, and extracellular matrices, with an algorithm that automatically polarizes cells. Both tools are highly efficient, enabling simulations of hundreds of thousands of deformable epithelial cells with unprecedented spatial fidelity. In this talk, I will demonstrate how these models reproduce key epithelial features and support applications including cancer growth, cell migration, and tissue stratification dynamics.
4. Eran Agmon University of Connecticut, USA
   "Multicellular Schema with Vivarium"
As models of multicellular biology grow in complexity, there is a critical need for modular, extensible frameworks that can manage diverse biological processes across scales. This talk introduces a compositional schema approach, a methodology for constructing simulations of multicellular systems through modular, interoperable components. Central to this approach is Vivarium, an open-source software platform designed to integrate heterogeneous modeling formalisms—including ODEs, stochastic kinetics, constraint-based models, agent-based systems, and rule-based logic—into unified, hybrid simulations. I will highlight how Vivarium supports schema-driven composition of cells, their interactions, and their environments, enabling scalable simulations of tissues, microbial communities, and synthetic consortia. A flagship example is the integration of diverse mechanistic submodels in a whole-cell simulation of Escherichia coli, demonstrating Vivarium’s ability to orchestrate biological complexity through standardized schemas and modular interfaces. I will discuss design principles behind Vivarium, share emerging tools for building multicellular models, and outline future directions for collaborative, open-ended systems biology.

Organized by: Daniel Glazar (Moffitt Cancer Center & Research Institute), Renee Brady-Nicholls, Moffitt Cancer Center & Research Institute

1. Franz Kuchling Allen Discovery Center, Tufts University
   "Uncertainty Minimization as an Adaptive and Evolutionary Imperative in Biology"
Recent advances in molecular biology have enabled precise manipulation of signaling pathways in living organisms, yet a unifying framework for predicting the organismal-level emergence of form and function remains elusive. The free energy principle, originally developed for neuroscience, offers a Bayesian inference approach to model cellular decision-making during morphogenesis and emergent aneural behavior. Simulations demonstrate the utility of this framework in explaining developmental anomalies (e.g., planarian axial polarity defects) and early carcinogenesis as consequences of maladaptive cellular 'beliefs.' Complementing this, evolutionary metacognition theory formalizes adaptation across timescales, illustrating how coevolutionary processes naturally favor the emergence of multi-scale regulatory systems. These metacognitive architectures promote energy-efficient responses to fluctuating selection pressures. Experimental observations in aneural systems such as Volvox algae support these predictions: Volvox colonies display adaptive phototaxis and retain stimulus-associated behaviors beyond exposure, suggesting a primitive form of memory. These insights offer a cross-disciplinary framework—integrating developmental biology, evolutionary theory, and basal cognition—to model adaptive behavior across biological scales. This foundation may inform future directions in modeling complex diseases such as cancer, particularly where cell-state decisions and misregulation mirror maladaptive inference processes.
2. Nathanaël Hozé Université Paris Cité, INSERM, IAME, F-75018, Paris, France
   "A multi-scale modelling framework to assess the relationship between SARS-CoV-2 viral load and transmission in household studies"
Understanding the drivers of SARS-CoV-2 transmission is essential for designing effective interventions, particularly in close-contact settings such as households. While viral load is widely believed to influence infectiousness, quantifying its role remains challenging due to individual variability, asymptomatic infections, and the unobservability of transmission events. Household studies offer a controlled context for investigating the link between viral load dynamics and transmission, especially when combined with high-frequency sampling. However, such designs are costly, and their added value relative to simpler approaches is unclear. We present a multi-scale modelling framework that integrates within-host viral dynamics and between-host transmission processes in household settings. We developed a stochastic agent-based model of viral dynamics that includes inter-individual variability. We developed a Bayesian inference approach implemented in rstan, in which we jointly estimate individual-level parameters, infection times, and the relation between viral load and transmissibilty. Our simulation-based framework evaluates whether monitoring viral load at high temporal resolution improves the reconstruction of transmission chains and the estimation of key epidemiological parameters. We compare this rich sampling design to two more commonly used alternatives: (i) designs based solely on symptom onset, and (ii) designs based on qualitative viral detection (i.e., positive/negative status without quantification). We show that incorporating quantitative viral load data improves the accuracy of transmission chain reconstruction and enhances the estimation of key metrics, including the probability of infection, generation interval, and incubation period. This work provides quantitative insights into the potential benefits of incorporating viral load measurements into household transmission studies and informs the design of future studies aimed at elucidating the role of viral kinetics in infectious disease spread.
3. Kathleen Wilkie Toronto Metropolitan University
   "Practical Parameter Identifiability and Handling of Censored Data with Bayesian Inference in Models of Tumour Growth"
Mechanistic mathematical models are a powerful tool to help us understand and predict the dynamics of tumour growth under various conditions. In this work, we use five models with an increasing number of parameters to explore how certain (often overlooked) decisions in estimating parameters from data affect the outcome of the analysis. In particular, we propose a framework for including tumour volume measurements that fall outside the upper and lower limits of detection, which are normally discarded. We demonstrate how excluding censored data results in an overestimation of the initial tumour volume and the model-predicted tumour volumes prior to the first measurements, and an underestimation of the carrying capacity and the predicted volumes beyond the latest measurable time points. We show how the choice of prior for the model parameters can impact the posterior distributions, and illustrate that reporting the most likely parameters and their 95% credible interval can lead to confusing or misleading interpretations. We hope this work will encourage others to carefully consider the choices made in parameter estimation and to consider adopting the approaches discussed in this talk.
4. Ernesto A. B. F. Lima The University of Texas at Austin
   "Modeling tumor sensitivity and resistance: a bayesian framework for predicting combination therapies"
Understanding the heterogeneous response of tumors to therapy remains a major challenge in oncology, particularly in the presence of treatment resistance. We present a framework to model the growth dynamics of radiation-sensitive and radiation-resistant breast cancer cells receiving radiotherapy, immunotherapy, and their combination. Using experimental data from murine models, we construct a family of ordinary differential equation models and apply Bayesian calibration and model selection to identify the most parsimonious model capable of capturing and predicting the observed experimental dynamics. Our approach quantifies differences between sensitive and resistant tumors. Resistant cells exhibited not only faster intrinsic growth rates but also a greater capacity for post-radiotherapy repair compared to sensitive cells. These biological differences were incorporated into the modeling through group-specific parameters, selected using the Bayesian Information Criterion to balance model complexity and predictive ability. In the immunotherapy arm, a pronounced heterogeneity in treatment response was observed. By performing mouse-specific calibration of key parameters governing immunotherapy efficacy and linking them to imaging-derived biomarkers, we successfully captured this variability across subjects. For the combination therapy predictions, the concordance correlation coefficient and Pearson correlation coefficient increased from 0.31 and 0.34 (without biomarkers) to 0.34 and 0.54 (with biomarkers), demonstrating the added predictive value of imaging-informed modeling. The Bayesian framework enabled robust parameter estimation, uncertainty quantification, and assessment of model identifiability, providing insights into the dynamics of combination therapy. Our results emphasize the importance of accounting for intra-tumoral heterogeneity in predictive modeling to improve treatment planning and evaluation.

Organized by: Jennifer Flegg (University of Melbourne), Prof Mat Simpson, Queensland University of Technology

Note: this minisymposia has multiple sessions. The other session is: and Part-2.

1. Ruth Baker University of Oxford
   "Optimal experimental design for parameter estimation in the presence of observation noise"
Using mathematical models to assist in the interpretation of experiments is becoming increasingly important in research across applied mathematics, in particular in fields such as biology and ecology. In this context, accurate parameter estimation is crucial; model parameters are used to both quantify observed behaviour, characterise behaviours that cannot be directly measured and make quantitative predictions. The extent to which parameter estimates are constrained by the quality and quantity of available data is known as parameter identifiability, and it is widely understood that for many dynamical models the uncertainty in parameter estimates can vary over orders of magnitude as the time points at which data are collected are varied. In this talk I will outline recent research that uses both local and global sensitivity measures within an optimisation algorithm to determine the observation times that give rise to the lowest uncertainty in parameter estimates. Applying the framework to models in which the observation noise is both correlated and uncorrelated demonstrates that correlations in observation noise can significantly impact the optimal time points for observing a system, and highlights that proper consideration of observation noise should be a crucial part of the experimental design process.
2. Yong See Foo University of Melbourne
   "Quantifying structural uncertainty in chemical reaction network inference"
Dynamical systems in biochemistry are complex, and one often does not have comprehensive knowledge about the interactions involved. Chemical reaction network (CRN) inference aims to identify, from observing species concentrations, the unknown reactions between the species. Most approaches focus on identifying a single, most likely CRN, without addressing uncertainty about the resulting network structure. However, it is important to quantify structural uncertainty to have confidence in our inference and predictions. To this end, I will discuss how to construct posterior distributions over CRN structures. This is done by keeping a large set of suboptimal solutions found in an optimisation framework with sparse regularisation, in contrast to existing optimisation approaches which discard suboptimal solutions. I will show that inducing reaction sparsity with nonconvex penalty functions results in more parsimonious CRNs compared to the popular lasso regularisation. In a real-data example where multiple CRNs have been previously proposed, reactions proposed from different literature can be simultaneously recovered under structural uncertainty. Moreover, posterior correlations between reactions help identify where structural ambiguities are present. This can be translated into alternative reaction pathways suggested by the available data, which guide the efforts of future experimental design.
3. Michael Pan The University of Melbourne
   "Thermodynamic modelling of membrane transport processes using bond graphs"
Cellular systems are physical systems, and are therefore governed by the laws of physics and thermodynamics. Energy is fundamental to our understanding of membrane transporters, which will only operate in the direction of decreasing chemical potential. Despite this, energy is often ignored in mathematical models of transporters, leading to unrealistic behaviours analogous to perpetual motion machines. In this talk, we outline a general physics-based framework (the bond graph) that explicitly models energy and therefore inherently accounts for thermodynamic constraints in membrane transporters. We show that this framework also provides a natural means of modelling the voltage dependence of electrogenic transporters. We demonstrate the utility of the bond graph approach in modelling the cardiac Na+/K+ ATPase (sodium-potassium pump) and discuss potential extensions of this approach for whole-cell modelling.
4. Jean (Jiayu) Wen The Australian National University
   "Advancing Genomic Foundation Models with Electra-Style Pretraining: Efficient and Interpretable Insights into Gene Regulation"
Pre-training large language models on genomic sequences has emerged as a powerful strategy for capturing biologically meaningful representations. While masked language modeling (MLM)-based methods, such as DNABERT and Nucleotide Transformer, achieve strong performance, they are hindered by inefficiencies due to partial token supervision and high computational demands. To address these limitations, we introduce the first Electra-style pretraining framework for genomic foundation models, replacing the MLM objective with a replaced-token detection task that employs a discriminator network to distinguish tokens replaced by a generator, enabling dense token-level supervision and significantly accelerating training. Unlike conventional methods that tokenize genomic sequences into 6-mers, our model operates at single nucleotide resolution, enhancing both efficiency and interpretability. We pre-train our model on the human genome and fine-tune it across a spectrum of downstream genomic prediction tasks, spanning epigenetics, transcriptional regulation, and post-transcriptional processes, including identification of regulatory elements such as promoters and enhancers, prediction of histone modifications, assessment of chromatin accessibility, as well as prediction of RNA-protein interactions, RNA modifications, RNA stability, translational efficiency, and microRNA binding sites. By addressing these diverse tasks, our model contributes to the advancement of whole cell modeling, which requires an integrated understanding of genomic, transcriptomic, and proteomic interactions. Our approach achieves a 28-fold reduction in pretraining time compared to MLM-based methods while surpassing their performance in most downstream evaluations, with benchmarking against state-of-the-art genomic models. Comprehensive ablation studies illuminate the key factors driving this improved efficiency and effectiveness. Furthermore, the use of 1-mer tokenization allows for nucleotide-level resolution, greatly enhancing the model's interpretability, with visualization and attention analyses demonstrating its ability to capture biologically relevant sequence motifs at a fine-grained level, providing deeper insights into genomic regulatory mechanisms. This work underscores the potential of Electra-style pretraining as a computationally efficient and effective strategy for advancing genomic representation learning, with broad implications for systems biology and whole cell modeling.

Organized by: Kelsey Gasior (University of Notre Dame)

1. Samuel Oliver Swansea University
   "The role of EMT in Ovarian Cancer: Insights from a Mathematical Model"
The role of EMT in Ovarian Cancer: Insights from a Mathematical Model Epithelial-to-mesenchymal transition (EMT) is a critical process in cancer progression that can significantly reduce the effectiveness of treatments. EMT occurs when cells undergo phenotypic changes, resulting in altered behaviours compared to their original state. This transition may lead to increased drug resistance, greater cell plasticity, and enhanced metastatic potential. As a result, understanding and studying the role of EMT in tumour progression and treatment response is essential. In this study, we utilise a 3D agent-based multiscale modelling framework with PhysiCell to examine the role of EMT over time in two ovarian cancer cell lines, OVCAR-3 and SKOV-3. This approach enables us to investigate the spatiotemporal dynamics of ovarian cancer and provide insights into the development of the tumours. The model incorporates microenvironmental conditions, adjusting cellular behaviours based on factors such as substrate concentrations and the proximity of neighbouring cells. The OVCAR-3 and SKOV-3 cell lines exhibit significantly different tumour architectures, allowing for the exploration of various tumour dynamics and morphologies. The model successfully captures biological patterns observed in tumour growth and progression, offering valuable insights into the dynamics of these cell lines. Additionally, sensitivity analysis is conducted to evaluate the impact of parameter variations on model outcomes.
2. Nate Kornetzke University of New Mexico
   "Turn down that noise! Uncertainty quantification for stochastic models of emerging infectious pathogens"
Emerging infectious pathogens are a persistent public health threat that challenge traditional mechanistic modeling approaches. As outbreaks initially start with a low number of infected hosts, the dynamics of these outbreaks are highly stochastic, making traditional deterministic methods, e.g. ordinary differential equations, unable to qualitatively or quantitatively capture the transmission dynamics. Instead, stochastic models are used, such as Markov chain models, but these models present their own challenges. Often, to infer a quantity of interest with these stochastic models, we need to sample the model’s distribution many times over, introducing an additional source of noise to our analysis. This additional noise can be amplified around bifurcating points of the model, making the inference of our quantity of interest even more difficult. Here, we show how novel tools from the field of uncertainty quantification can be used to disentangle noise in stochastic systems to make rigorous statistical inferences that are crucial for modeling emerging pathogens. We illustrate these techniques with a model of yellow fever virus spillover in the Americas, a virus that has seen rapid emergence amongst multiple hosts and vectors in South America over the last decade.
3. Steve Williams University of California, Merced
   "Examining models of phenotype selection in populations of bacteria under external predatory stress"
Biofilms are dense communities of bacteria living in a collective extracellular matrix. They aid their constituent bacteria by protecting them from external environmental threats, distributing metabolic workload, and performing complex multicellular processes (e.g., quorum sensing). However, for many organisms, retaining the tools necessary to be multiple phenotypes within their lifetime has been essential for survival. It is natural to wonder how external stressors in marine environments can impact the biofilm formation process and whether these impacts have downstream implications for their participation in multi-organism relationships. To explore this adaptation process, we have employed a previously proposed population model in which bacteria transition freely between planktonic and biofilm phenotypes in the presence of predator. By employing several sensitivity analysis techniques, we probe the parameter space to understand the impacts that changing bacterial dynamics can have. Using synthetic data, we have quantified uncertainties present in the realization of our system using practical identifiability techniques. Finally, we propose a new model with parameter variations within the bacterial population, particularly their ability to attach to biofilms. Through the lens of sensitivity analysis again, this model allows us to begin to measure the rate of adaptation for such populations in terms of the size of the external stressors and the distribution of the variation inside the population.
4. Kelsey Gasior University of Notre Dame
   "Comparative Sensitivity Analyses and Modeling the Epithelial Mesenchymal Transition"
The epithelial mesenchymal transition (EMT) is a process that allows carcinoma cells to lose their adhesivity and migrate away from a tumor. Further, cells can maintain this invasiveness after they leave their original microenvironment, suggesting that there is an underlying bistable switch. We developed a mathematical model that examined the relationships between E-cadherin and Slug and their responses to tumor-level factors, such as cell-cell contact and TGF-b. Phenomenological model behavior was derived from biological experiments and, ultimately, this model showed how cells at different positions within a tumor can use exogenous factors to undergo EMT. However, the nonlinear dynamics and estimated model parameters make it challenging to analyze and understand what parameters contribute to the observed E-cadherin and Slug changes. Thus, we turn to sensitivity analysis. This work seeks to understand the true impact of mathematical and statistical techniques on our understanding of the dynamics underlying EMT. To provide an extensive understanding, multiple ranges were examined for each parameter and techniques such as nondimensionalization, Latin Hypercube Sampling, Partial Rank Correlation Coefficient, Morris Method Screening, and Sobol’ analysis were used. This wide range of techniques was applied to cells exposed to different levels of cell-cell contact and exogenous TGF-b. By comparing these different methodologies, parameter ranges, and treatment groups, a dual biological and mathematical perspectives emerge. While the different analytical techniques highlight different parameters of importance and interacting relationships, comparing across treatment groups shows how a cell’s identity can be controlled by different intracellular factors, which may shift in the dynamic tumor environment. Together, these results highlight the need for extensive and methodical approach to sensitivity analysis before conclusions can be reached to inform future experiments.

Organized by: Jinsu Kim (POSTECH), Eric Foxall (The University of British Columbia - Okanagan Campus), and Linh Huynh (Dartmouth College)

Note: this minisymposia has multiple sessions. The other sessions are: Part-2, Part-3, and Part-4.

1. Jinsu Kim POSTECH
   "Stability of stochastic biochemical reaction networks"
A reaction network is a graphical representation of interactions between chemical species (molecules). When the species' abundances in the system are low, the inherent randomness of molecular interactions significantly influences the system dynamics. In such cases, the abundances are modeled stochastically using a continuous-time Markov chain that evolves in a jump-by-jump fashion. A major challenge in this area is establishing the stability of the Markov chain—that is, proving the existence of a stationary distribution. Another goal is to find a closed form of the stationary distribution, which is often extremely difficult. In this talk, I will present structural conditions on reaction networks that guarantee stability. I will also introduce novel techniques, inspired by reaction network theory, that aid in deriving closed-form expressions for stationary distributions.
2. Daniel Schultz Dartmouth College
   "Emergence of heterogeneity during bacterial antibiotic responses"
Heterogeneity is a fundamental aspect of microbial ecology, conferring resilience and adaptability to microbial populations. Remarkably, this heterogeneity does not necessarily depend on complex mechanisms of differentiation into specialized phenotypes but can instead be achieved through fundamental processes that are common to all microbes. Yet, despite significant advances in describing microbial physiology, we still lack a framework to bridge across scales to connect cellular processes to emergent behaviors in microbial communities. Here, we aim to understand how stochastic and spatial variations affect cellular metabolism and thereby provide a mechanism to generate phenotypic diversity, giving rise to complex collective behaviors in microbial populations. To overcome difficulties in studying cell responses across multiple scales, we use single-cell and biofilm microfluidics to develop mathematical models of antibiotic response dynamics. Our single-cell microfluidic experiments capture a remarkable variety of phenotypes caused by stochastic fluctuations during drug responses. Similarly, our model predicts the coexistence of stable phenotypes corresponding to growing and arrested cells. We describe the nature and stability of these different phenotypes and connect single-cell heterogeneity to population-level growth. In our biofilm microfluidic experiments, the formation of nutrient gradients due to spatial variations across the population results in a range of metabolic states with different antibiotic susceptibilities. Drug exposures result in a major reorganization of the biofilm, giving rise to collective behaviors such as increased resistance and “memory” from past exposures. A spatial version of our model describes the contribution of spatial structure to the collective mechanisms of resistance provided by organization into biofilms, showing how spatially structured populations can survive much higher drug doses than planktonic populations. Together, these results elucidate how heterogeneity emerges in microbial populations and how it gives rise to complex behaviors at the population level.
3. Anna Kraut St. Olaf College
   "Evolution across fitness valleys in a changing environment"
The biological theory of adaptive dynamics aims to study the interplay between ecology and evolution under the basic mechanisms of heredity, mutation, and competition. The typical evolutionary behavior of a population can be studied mathematically by looking at macroscopic limits of large populations and rare mutations derived from microscopic individual-based Markov processes. Previous work has been focused on a variety of scaling regimes and the resulting stochastic and deterministic limit processes under the assumption of constant environmental parameters. In our present work, we relax this assumption and study repeating changes in the environment, allowing for all of the model parameters to vary over time as periodic functions on an intermediate time scale between those of stabilization of the resident population (fast) and exponential growth of mutants (slow). Biologically, this can for example be interpreted as the influence of seasons or the fluctuation of drug concentration during medical treatment. Analyzing the influence of the changing environment carefully on each time scale, we are able to determine the effective growth rates of emergent mutants and their invasion of the resident population. In recent work, we study the crossing of so-called fitness valleys, where multiple disadvantageous mutations need to be accumulated to gain a fitness advantage. A changing environment has interesting implications for the crossing rates of such valleys, particularly if some of the intermediate traits occasionally have positive growth rates and can serve as pit stops.
4. Eric Foxall UBC Okanagan
   "Perturbation theory of reproductive value for branching Markov processes"
The reproductive value gives an individual’s relative contribution to the success of a population, as a function of its type, and for suitably recurrent models can be computed via a renewal argument. We show that the same argument can be used to compute its sensitivity in terms of an associated Markov process that describes the trajectory of a distinguished lineage, known to probabilists as the spinal particle. In the case of age-structured models this leads to nice formulas for the dependence of reproductive value on parameters.

Organized by: Tomas Helikar (University of Nebraska - Lincoln), Juilee Thakar (Juilee_Thakar@URMC.Rochester.edu) - University of Rochester Medical Center James Glazier (jaglazier@gmail.com) - Indiana University

Note: this minisymposia has multiple sessions. The other sessions are: Part-2, and Part-3.

1. Gary An University of Vermont
   "NASEM-compliant Critical Illness Digital Twins to cure sepsis"
To date there are no pharmacological agents that can substantively and reliably affect the underlying host pathophysiology of sepsis. The effective control of sepsis requires personalized precision medicine, which requires the capabilities provided by digital twins compliant with industrial standards and consistent with the definition put forth in the National Academies of Science, Engineering and Medicine (NASEM) report entitled 'Foundational Research Gaps and Future Directions for Digital Twins' that provides an operational definition for a digital twin and lists specific challenges moving forward for the development of this technology. NASEM defines a digital twin thusly: 'The key elements that comprise a digital twin include (1) modeling and simulation to create a virtual representation of a physical counterpart, and (2) a bidirectional interaction between the virtual and the physical. This bidirectional interaction forms a feedback loop that comprises dynamic data-driven model updating (e.g., sensor fusion, inversion, data assimilation) and optimal decision-making (e.g., control, sensor steering).' Notably, this definition is not met by the vast majority of currently described biomedical “digital twins,” and this insufficiency limits the applicability of non-NASEM compliant digital twins to provide the true personalized precision medicine required to treat complex immune diseases such as sepsis. We present a prototype Critical Illness Digital Twin developed with a workflow that utilizes mechanistic models with machine learning and artificial intelligence for clinically relevant parameter space identification, trajectory personalization, discovery of novel multimodal/adaptive therapeutic control and guidance for sensor/actuator development. The CIDT is based on a previously validated agent-based model of systemic inflammation, and constructed to conform to a mathematical object terms the Model Rule Matrix (MRM). The MRM employs the Maximal Entropy Principle to account for the latent space of 'what is left out' (e.g. Epistemic Uncertainty) in the rule structure of the CIDT. Operating on the CIDT with a workflow that includes genetic algorithms and active learning we identified non-falsifiable configurations of the MRM with respect to two distinct clinical cytokine time-series datasets, one for burns, one for trauma. We further applied deep reinforcement learning to train an artificial intelligence that can cure sepsis arising from a novel pathogen by modulating host cytokines using only currently FDA-approved biologics. Additional future work must include testing with a sufficiently complex large animal model that can recapitulate the heterogeneity seen in clinical sepsis.
2. Yi Jiang Georgia State University
   "Immunogenic Cell Death: The Key to Unlocking the Potential for Combined Radiation and Immunotherapy"
Immunogenic cell death (ICD) enhances anti-tumor immunity by releasing tumor-associated antigens and activating the anti-tumor immune system response. Here, we develop a mathe- matical model to quantify the role of ICD in optimizing the efficacy of combined radiotherapy (RT) and macrophage-based immunotherapy. Using preclinical murine data targeting the SIRPα-CD47 checkpoint, we show that RT alone induces minimal ICD, whereas disrupting the SIRPα-CD47 axis significantly enhances both phagocytosis and systemic immune activation. Our model predicts an optimal RT dose (6–8 Gy) for maximizing ICD, a dose-dependent abscopal effect, and a hierarchy of treatment efficacy, with SIRPα-knockout macrophages exhibiting the strongest tumoricidal activity. These findings provide a quantitative framework for designing more effective combination therapies, leveraging ICD to enhance immune checkpoint inhibition and radiotherapy synergy.
3. Hana Dobrovolny Texas Christian University
   "Incorporating the immune response into models of oncolytic virus treatment of cancer"
Oncolytic viruses present a promising path for cancer treatment due to their selectivity in infecting and lysing tumor cells and their ability to stimulate the immune response. While the immune response can help eliminate the tumor, it also acts to clear the virus and often limits the effectiveness of oncolytic virus therapy. Using experimental data, we test models of oncolytic virus infections incorporating various immune components in order to determine the most suitable immune models. We use the models to investigate the role of the immune response in oncolytic virus treatment, finding that a moderate immune response can prolong the oncolytic virus infection, allowing the virus to infect and kill more tumor cells than either a weak or strong immune response.
4. Jason E. Shoemaker University of Pittsburgh
   "Network representation of sex-specific immunity: A steppingstone to digital twins?"
In a world of immense and growing computational power, the eventual rise of Digital Twins will enable a degree of personal health optimization that is currently unimaginable. There are important questions on how society gets there, the ethics of owning one’s digital twin, and many more important questions to address as we progress towards the Digital Twin world. One small but important question in the short term is how we can use currently available tools to design personal treatments today or guide drug discovery. In our lab, we have leaned heavily on using molecular interaction networks as baseline models of human gene regulation. We have both independently and with colleagues developed new algorithms that can integrate interaction data and gene expression data to predict either drug mechanisms of action or pathways for suppressing respiratory virus replication. Now, we are using these tools to explore for antiviral drug targets that are sex-specific, meaning proteins that, when targeted, may help regulate virus replication specially in male or females. And we are extending these studies to determine what roles hormones may play as well. Here, we will discuss our early results wherein we have analyzed primary human nasal cells from male and female donors. Our early results show that network-based representations of gene regulation better isolate hormone regulated pathways, including inflammation pathways important to respiratory infection. With sufficient data, network-based approaches combined with machine learning may be a promising approach developing early digital twins that are relevant to respiratory infection.

Organized by: Rebecca Crossley, Stéphanie Abo (University of Oxford), University of Oxford

1. John Nardini The College of New Jersey
   "Decoding agent-based model behavior: novel methods for prediction and global sensitivity analysis"
Agent-based models (ABMs) are invaluable tools for studying the emergence of collective behavior in biology. Unfortunately, it is challenging to analyze ABM behavior due to their computational and stochastic nature. In this talk, I will present two recent studies aimed at developing new methodologies to enable the prediction, interpretation, and analysis of ABMs. In the first study, we use biologically-informed neural networks (BINNs) to forecast and predict ABM behavior. In particular, we show BINNs can learn interpretable differential equations to predict ABM data at new parameter values, and demonstrate this success using three case study ABMs of collective migration. In the second study, we combine several machine learning algorithms to develop a global sensitivity analysis pipeline for ABMs that is capable of identifying sensitive parameters, revealing common model patterns, and linking input model parameters to these patterns using a spatial ABM of tumor spheroid growth. Taken together, these studies demonstrate how concepts from machine learning are valuable for studying ABMs and will advance data-driven ABM modeling.
2. Jinchao Feng Great Bay University
   "A Sparse Bayesian Learning Algorithm for Estimation of Interaction Kernels in Motsch-Tadmor Model"
In this talk, we present a data-driven framework for identifying asymmetric interaction kernels in the Motsch–Tadmor model based on observed agent trajectories. Unlike symmetric models, the asymmetric setting introduces a nonlinear inverse problem due to the normalization of interaction weights. We reformulate the problem using the implicit form of the governing equations, reducing kernel learning to a subspace identification task. To solve this, we develop a sparse Bayesian learning approach that incorporates prior structure and quantifies uncertainty, enabling robust model selection under noise. Numerical experiments on several prototype systems demonstrate the method's ability to recover key interaction patterns and predict collective behavior accurately, even with limited or noisy data.
3. Seungwoong Ha Santa Fe Institute
   "Toward a Data-Centric Understanding of Collective Dynamics"
Understanding how collective behavior emerges from local interactions is a central question in modeling biological systems. While traditional approaches often assume fixed interaction rules, recent advances in data-driven modeling offer ways to infer these laws directly from empirical observations. In this talk, I present a set of machine learning-based methods developed to recover interaction structures and underlying dynamics in complex systems, from physical to population-level collective behavior. Across different scenarios, these approaches infer continuous interaction strengths, capture emergent phenomena not present in training data, and remain applicable to stochastic or temporally evolving systems. I will also highlight how adaptive agents can develop robust coordination strategies through learning in uncertain environments. These results suggest new pathways for linking observed dynamics to latent interaction rules, offering complementary tools to classical models of population dynamics.
4. Ming Guo Massachusetts Institute of Technology
   "Collective curvature sensing and fluidity in three-dimensional multicellular systems"
Collective cell migration is an essential process throughout the lives of multicellular organisms, for example in embryonic development, wound healing and tumour metastasis. Substrates or interfaces associated with these processes are typically curved, with radii of curvature comparable to many cell lengths. Using both artificial geometries and lung alveolospheres derived from human induced pluripotent stem cells, here we show that cells sense multicellular-scale curvature and that it plays a role in regulating collective cell migration. As the curvature of a monolayer increases, cells reduce their collectivity and the multicellular flow field becomes more dynamic. Furthermore, hexagonally shaped cells tend to aggregate in solid-like clusters surrounded by non-hexagonal cells that act as a background fluid. We propose that cells naturally form hexagonally organized clusters to minimize free energy, and the size of these clusters is limited by a bending energy penalty. We observe that cluster size grows linearly as sphere radius increases, which further stabilizes the multicellular flow field and increases cell collectivity. As a result, increasing curvature tends to promote the fluidity in multicellular monolayer. Together, these findings highlight the potential for a fundamental role of curvature in regulating both spatial and temporal characteristics of three-dimensional multicellular systems.

Organized by: Kang-Ling Liao (University of Manitoba), Wenrui Hao, Pennsylvania State University

Note: this minisymposia has multiple sessions. The other session is: and Part-1.

1. Kang-Ling Liao University of Manitoba
   "Mathematical Modeling of Breast Cancer Treatment with Radiation, Anti-estrogen, and Immune Checkpoint Inhibitor"
Radiotherapy (RT) and endocrine therapy (ET) are standard treatments for estrogen receptor-positive (ER+) breast cancer, but they could induce resistance and relapse issues. Immune checkpoint inhibitor (ICI) is another potential treatment for breast cancer, but its response rate is low. In this work, we create a system of ordinary differential equations to investigate the combination treatments among RT, ET, and ICI in ER+ breast cancer. Our model quantitatively captures the tumor growth data under the combination among these three treatments for different ER+ breast cancer cell lines. Our numerical predictions indicate that: (i) potential treatment to reduce the relapse caused by RT; (ii) potential breast cell lines have a better response rate to anti-PD-1; (iii) Tumor elimination and no relapse could appear in the combination of RT and ET in MCF-7 ER+ tumor cells; (iv) these treatments have induce better tumor reduction is which breast cancer cell lines. We also study the distribution of parameter values calibrating to different ER+ breast cancer cell lines to categorize (virtual) cohort patients and to provide potential biomarkers for selecting appropriate treatment for patients
2. Tracy Stepien University of Florida
   "Modeling Tumor-Immune Interactions in the Glioblastoma Microenvironment"
Glioblastoma (GBM) is an aggressive brain tumor that is extremely fatal with no current treatment options available that can achieve remission. One potential explanation for minimally effective treatments is due to the characteristically high immune-suppressive glioma microenvironment. We develop an agent-based model to simulate the interactions of glioma cells, T cells, and myeloid-derived suppressor cells (MDSCs) and the effects of oxygen, a T cell chemoattractant, and an MDSC chemoattractant. To validate our model and quantify cell clustering patterns in GBM, we use spatial statistics comparing simulations to data extracted from cross-sectional tumor images of cellular biomarkers.
3. Wenrui Hao Pennsylvania State University
   "Data-Driven Modeling in Alzheimer's Disease"
Alzheimer’s disease (AD) presents complex, nonlinear progression patterns driven by heterogeneous biomarker dynamics, spatial brain changes, and individual variability. Capturing these intricate dynamics requires modeling tools that are both expressive and computationally efficient. In this work, we introduce a Laplacian Eigenfunctions Neural Operator (LENO) framework for data-driven modeling of Alzheimer’s disease. By projecting the underlying spatial-temporal dynamics onto a basis of Laplacian eigenfunctions, LENO exploits the geometric structure of the brain while learning the nonlinear operator governing disease progression. Trained on multi-modal longitudinal data—including neuroimaging, biomarker profiles, and cognitive assessments—our approach enables efficient approximation of disease trajectories and identification of multiple progression pathways. The model not only achieves high predictive accuracy but also reveals interpretable spatial patterns aligned with known AD pathology. LENO provides a powerful computational tool for building digital twins of AD patients, supporting personalized diagnosis, subtyping, and forecasting of disease evolution.
4. Negar Mohammadnejad University of Alberta
   "Strategies for Optimizing the Efficacy of Oncolytic Virus–Immune System Interactions"
Oncolytic virotherapy (OVT) is an innovative cancer treatment in which oncolytic viruses are introduced into a patient to selectively target and destroy tumor cells. In the absence of these viruses, tumors are known to create an immunosuppressive environment. However, upon administration of oncolytic viruses and initiation of virotherapy, the immune system is activated, leading to a robust anti-tumor response. Despite this, oncolytic viruses alone have rarely been shown to induce complete and sustained regression of established tumors in vivo. In this talk, I will discuss key strategies for enhancing the efficacy of oncolytic virotherapy. These include the integration of immunotherapy approaches with virotherapy to amplify anti-tumor immune responses, as well as optimizing the timing, dosage, and sequencing of viral administrations to maximize therapeutic benefits. By refining these strategies, we aim to improve treatment outcomes and potentially enhance the therapeutic impact of oncolytic virotherapy.

Organized by: Sydney Ackermann, Ramanarayanan Kizhuttil, Samrat Sohel Mondal (Wodarz lab) (University of California, San Diego)

Note: this minisymposia has been accepted, but the abstracts have not yet been finalized.

Organized by: Dhananjay Bhaskar (Yale University), Bernadette Stolz-Pretzer

Note: this minisymposia has multiple sessions. The other session is: and Part-2.

1. Katherine Benjamin University of Oxford
   "Topological methods for subcellular spatial transcriptomics"
Spatial transcriptomics technologies produce gene expression measurements at millions of locations across a tissue sample. An open problem in this area is the inference of spatial information about single cells. Here we present a multiscale machine learning method to pinpoint the locations of individual sparsely dispersed cells from subcellular spatial transcriptomics data. We integrate this approach with multiparameter persistence landscapes, a state of the art tool in topological data analysis, to identify a loop structure in infiltrating glomerular immune cells in a mouse model of lupus nephritis.
2. Veronica Ciocanel Duke University
   "Unraveling aster and ring structures in cell models of dynamic actin filaments using topological data analysis"
Actomyosin is a dynamic network of interacting proteins that reshapes and organizes in a variety of structures that are essential in cell movement, cell division, as well as in wound healing. Agent-based models can simulate realistic dynamic interactions between actin filaments and myosin motor proteins inside cells. These stochastic simulations reproduce bundles, clusters, and contractile rings that resemble biological observations. We have developed techniques based on topological data analysis to extract insights from spatio-temporal data in these protein network interactions. Recently, we have been interested in adapting the framework of vines and vineyards in order to track topological and geometrical features through time-parameterized stacks of persistence diagrams. This approach allows us to quantify characteristics of formation and maintenance of relevant actin structures such as rings and asters in simulated datasets. This is joint work with Niny Arcila-Maya.
3. Robert McDonald University of Oxford
   "Topological model selection: a case-study in tumour-induced angiogenesis"
Comparing mathematical models offers a means to evaluate competing scientific theories. However, exact methods of model calibration are not applicable to many probabilistic models which simulate high-dimensional spatio-temporal data. Approximate Bayesian Computation is a widely-used method for parameter inference and model selection in such scenarios, and it may be combined with Topological Data Analysis to study models which simulate data with fine spatial structure. We develop a flexible pipeline for parameter inference and model selection in spatio-temporal models. Our pipeline identifies topological summary statistics which quantify spatio-temporal data and uses them to approximate parameter and model posterior distributions. We validate our pipeline on models of tumour-induced angiogenesis, inferring four parameters in three established models and identifying the correct model in synthetic test-cases.
4. Nan Wu University of Texas at Dallas
   "Adaptive Bayesian regression on manifold"
We investigate how the posterior contraction rate under a Gaussian process prior is influenced by the intrinsic dimension of the domain and the smoothness of the regression function. Specifically, we consider the setting where the domain is a d-dimensional manifold and the regression function is intrinsically s-Hölder smooth on the manifold. We establish the optimal posterior contraction rate of O(n^{-s/(2s + d)}), up to a logarithmic factor. To eliminate the need for prior knowledge of the manifold's dimension, we propose an empirical Bayes prior on the kernel bandwidth, leveraging kernel affinity and k-nearest neighbor statistics. This talk is based on joint work with Tao Tang, Xiuyuan Cheng, and David Dunson.

Organized by: Hye-Won Kang (University of Maryland Baltimore County), Arnab Ganguly, Louisiana State University, aganguly@lsu.edu

Note: this minisymposia has multiple sessions. The other sessions are: Part-2, and Part-3.

1. Ruth J Williams University of California San Diego
   "Stochastic Analysis of Markov Chain Models for Chromatin Dynamics"
Stochastic dynamics and time-scale differences between establishment and erasure processes in chromatin modifications (such as histone modifications and DNA methylation) have been seen in simulations to have a critical effect on maintaining and switching cell types through generations of cell division. It has been further observed that cross-catalysis between repressive histone modifications and DNA methylation can quickly silence a gene, and protein-mediated positive autoregulation can alleviate this silencing. In this talk, we provide a rigorous mathematical framework to validate, explain and extend these insights. We introduce stochastic models of chromatin modification circuits as singularly perturbed, continuous-time Markov chains with a small parameter epsilon capturing the time scale separation. We characterize the limiting stationary distribution as epsilon goes to zero in terms of a reduced Markov chain. We also show that protein-mediated positive autoregulation can monotonically alleviate cross-catalytic silencing caused by two types of repressive modifications. The theoretical tools developed not only provide a solid mathematical foundation for previous computational and experimental findings, emphasizing the role of chromatin modification dynamics and protein-mediated autoregulation, but also have broader applications to singularly perturbed continuous time Markov chains, particularly those associated with chemical reaction networks. Based on joint work with S. Bruno, Felipe Campos, D. Del Vecchio, Y. Fu.
2. Grzegorz Rempala Ohio State University
   "Likelihood Functions for Individual-Level Chemical Reaction Models"
When analyzing chemical reaction systems, it is often valuable to track the behavior of individual molecules over time. In such settings, one can construct an individual-level likelihood function—a statistical tool that quantifies how well a specific parametric reaction model explains observed data. Such likelihood functions are particularly useful when applied to time series data that capture the trajectories of chemical reaction networks. In this talk, I will introduce the concept of individual-level likelihoods, highlight their key applications, and discuss practical approximations, especially in the context of mass-transfer models. A central example will be the stochastic SIR model, though similar constructions apply more broadly across biological and chemical systems.
3. Mark Flegg Monash University
   "Stochastic Simulation of Reaction Networks with Well-Mixed Clustered Agents"
In this talk we will explore the suitability of simulating reaction networks at the level of local clusters rather than individuals for improvements in efficiency and reduction of complexity. This approach asserts an approximation with how a local cluster of reactants evolve and specifically how the components of this cluster interact with the larger network. We explore the method in the context of simple population models of a disease. Here, clusters represent the efficient disease interactions within households and approximations are made with how individuals of a household contribute to spreading the disease in the rest of the community. In biochemical systems a cluster is more complicated and constitutes efficient molecular mechanisms of multiple components embedded in a larger chemical network.
4. Hye-Won Kang University of Maryland Baltimore County
   "Multiscale Approximation and Parameter Estimation in Stochastic Models of the Glycolytic Pathway"
In this talk, I will introduce a glycolytic pathway that includes multiple enzyme-catalyzed reactions. We focus on the part involving the phosphofructokinase (PFK) reaction as a case study in stochastic modeling. Using model reduction techniques, we show how to derive a simplified model and use it to estimate parameters from partially observed data. Previous studies modeled this pathway deterministically and employed a quasi-steady-state approximation to reduce its complexity. In contrast, we assume that some enzymes are present in low copy numbers and thus adopt a continuous-time Markov chain framework to capture stochastic effects. To further reduce network complexity, we apply a multiscale approximation method and derive a reduced ODE model that describes the system's behavior on a slow timescale. The reduced model focuses on two key species: fructose-6-phosphate (F6P) and adenosine diphosphate (ADP). It not only captures the essential dynamics of the full network but also provides insights into key parameters. The equations in the reduced model contain fewer parameters—expressed as functions of those in the full model--which facilitates more tractable parameter estimation. Assuming that only the reduced species are observable, we generate synthetic data from the full model and use it to estimate the parameters in the reduced model. This approach demonstrates how time-series data from a subset of species can enable effective estimation of composite parameters in a reduced system. This is joint work with Arnab Ganguly.

Organized by: Cody FitzGerald (Northwestern University), Rainey Lyons (CU Boulder), Nora Heitzman-Breen (CU Boulder), Susan Rogowski (NCSU)

Note: this minisymposia has multiple sessions. The other sessions are: Part-1, Part-2, and Part-4.

Note: this minisymposia has been accepted, but the abstracts have not yet been finalized.

Organized by: Maria Kleshnina (Queensland University of Technology), Mason Lacy (Queensland University of Technology), Luke Filippini (Queensland University of Technology)

1. Luke Filippini Queensland University of Technology
   "Data-informed stochastic frameworks of anisotropic movement in the brain"
Neurological diseases and disorders are the subject of an extensive area of research that is of significant importance to the scientific community and wider population. Notable examples include autism, multiple sclerosis, and nervous system cancers, such as glioblastoma, which currently remain incurable. This is primarily due to the structural complexity of the nervous system and the impracticalities of surgical examination and/or resection. Hence, indirect methods, such as magnetic resonance imaging and mathematical modelling, are frequently relied upon to yield meaningful insight into the physiological processes that drive disease progression. In this talk, we discuss methods for deriving data-informed stochastic models from deterministic frameworks of anisotropic particle diffusion, motivated by applications to neurological diseases and disorders. We consider on-lattice stochastic models derived from a finite volume discretisation of the diffusion equation coupled with diffusion tensor imaging data. Furthermore, we discuss the limitations of using a traditional square or rectangular lattice, in terms of obtaining non-negative transition probabilities, and present a more promising approach using a hexagonal lattice. Most notably, the latter approach yields non-negative transition probabilities for any valid diffusion tensor.
2. Moriah Echlin Tampere University
   "Using Single-Cell Data-driven Boolean Network Models to Analyze Prostate Cancer Progression"
Cancer is a multifaceted disease, with many unique drivers; yet all cancers have a common foundation – the abnormal and malignant behavior of the body’s cells. Broadly, cellular behaviors result from the dynamics of the gene regulatory network (GRN) and genetic mutations can force the GRN into irregular dynamics. Thus, cells can exhibit the pathological properties associated with cancer: unchecked growth, immune evasion, and metastasis. To understand the origins and ramifications of malignant changes to the GRN, we combine clinically relevant single-cell transcriptomic data with a dynamical systems theoretical framework. This approach takes advantage of the system-wide gene correlations and cell state heterogeneity captured in single-cell ‘omics and the temporal and functional structure provided by dynamical systems models. Specifically, we use a Boolean network architecture to convert distinct cellular profiles to dynamical states. Our work focuses on the conversion of single-cell transcriptomic data to informative Boolean states and their subsequent analysis with the aim of identifying disease-relevant genes, inter-gene dependencies, and cell state dynamics that would not be evident in the original unstructured data. In particular, we highlight changes to the cell state structure that occur as cancer progresses from a primary indolent tumor to metastatic treatment-resistant disease.
3. Louise Spekking TU Delft
   "Improving cancer therapy through migrastatics and estimating tumor composition"
Adaptive therapy, which anticipates and forestalls the evolution of resistance in cancer cells, has gained significant traction, especially following the success of the Zhang et al.'s protocol in treating metastatic castrate-resistant prostate cancer. While several adaptive therapies have now advanced to clinical trials, none currently incorporates migrastatics, i.e. treatments designed to inhibit cancer cell metastasis. In this study, we propose the integration of migrastatics into adaptive therapy protocols and an evaluation of the potential benefits of employing a game-theoretic spatial model. Our results demonstrate that the combination of adaptive therapy with migrastatics effectively delays the onset of metastasis and reduces both the number and size of metastases across the majority of cancer scenarios . This approach not only extends the time to the first metastasis but also enhances the overall efficacy of adaptive therapies. Our findings suggest a promising new direction for cancer treatment, where adaptive therapy, in conjunction with migrastatic agents, can target both the evolution of resistance and the metastatic spread of cancer cells. In treatment of cancers, success strongly depends on our ability to capture how the disease evolves in response to treatment, both in terms of the size and composition. Understanding the changes in these compositions and the composition of the tumor will aid in developing new therapies in the future. In the second part of the talk, we will assess different machine learning methods on the deconvolution of cells based on microarray RNA sequencing data of glioblastoma organoids. Here, we show that the proportion of cell types changes over time with treatment and that these changes differ between organoids. We believe that this methodology can help in designing better therapies through testing evolutionary responses in patient-derived organoids, while in parallel the ecological response can be tracked through serum biomarkers and imaging in the corresponding patients. This will improve the adoption of adaptive therapies in clinical practice. Joint work with Jan Brábek, Joel S. Brown, Rachel Cavill, Robert A. Gatenby, Christopher Hubert, Weronika Jung, Christer Lohk, Barbora Peltanová, Daniel Rösel, Katharina Schneider, Maikel Verduin, Marc Vooijs, Sepinoud Azimi and Kateřina Staňková.
4. Noa Levi University of Melbourne
   "Leveraging algebraic approaches to inform therapeutic intervention"
The propensity for biological systems to exhibit adaptation is thought to play an important role in many treatment failures, especially in the context of cancer, since the underlying signalling networks under which cancer thrives are frequently able to adapt to the therapy. Here we present a general mathematical framework to study the effect of targeted pharmacological intervention in intracellular signalling networks which exhibit adaptation. This framework combines methods from graph theory and algebraic geometry to explain why treatment often fails, while illuminating alternative treatment strategies which may offer more success.

Organized by: Stuart Johnston (The University of Melbourne), Matthew Faria

1. Irina Kareva Northeastern University
   "From pre-clinical data to first in human dose projections: a different puzzle every time"
Identification and pre-clinical testing to determine dose-response relationships for new compounds is only one step in the drug discovery and development process. Translation of pre-clinical data into efficacious and safe human doses is a separate process, which may rely on assessing target occupancy, biomarker modulation, benchmarking against competitor compounds, and other detective work, to identify compound-specific criteria to predict safety and efficacy criteria in humans. In this talk, I will review the general approaches that help us make these predictions and then go through a few examples to describe how we solved the dose prediction puzzle for each individual case.
2. Yun Min Song Institute for Basic Science
   "Beyond FDA Guidance: Enhancing Accuracy in Predicting Drug-Drug Interactions"
The U.S. Food and Drug Administration (FDA) guidance recommends several model-based approaches to assess potential drug-drug interactions (DDIs) mediated by enzyme induction or inhibition. However, the FDA-recommended equation for predicting DDIs mediated by induction is known to have low accuracy. Moreover, a systematic and reliable method for accurately estimating inhibition constants—key parameters for predicting DDIs mediated by inhibition—remains lacking, with considerable inconsistencies reported across studies. In this talk, we demonstrate that the inaccuracy of the FDA-recommended equation stems from a fundamental limitation of the Michaelis-Menten equation, and we propose an improved model that considerably enhances predictive accuracy (~2-fold). Furthermore, we introduce a novel, optimized method for estimating inhibition constants, termed 50-BOA, which substantially reduces the number of required experiments (>75%)% while maintaining high precision and accuracy.
3. Stuart Johnston The University of Melbourne
   "Quantifying biological heterogeneity in nano-engineered particle-cell interaction experiments"
Nano-engineered particles are a promising tool for medical diagnostics, biomedical imaging, and targeted drug delivery. Fundamental to the assessment of particle performance are in vitro particle-cell interaction experiments. These experiments can be summarised with key parameters that facilitate objective comparisons across various cell and particle pairs, such as the particle-cell association rate. Previous studies often focus on point estimates of such parameters and neglect heterogeneity in routine measurements. In this study, we develop an ordinary differential equation-based mechanistic mathematical model that incorporates and exploits the heterogeneity in routine measurements. Connecting this model to data using approximate Bayesian computation parameter inference and prediction tools, we reveal the significant role of heterogeneity in parameters that characterise particle-cell interactions. We then generate predictions for key quantities, such as the time evolution of the number of particles per cell. Finally, by systematically exploring how the choice of experimental time points influences estimates of key quantities, we identify optimal experimental time points that maximise the information that is gained from particle-cell interaction experiments.
4. Thibault Delobel Institut Curie
   "Integrating glioblastoma plasticity into combination treatment design: a quantitative systems pharmacology and machine learning approach"
Glioblastoma (GBM) is the most aggressive primary brain tumor in adults, with a median survival below 18 months and no curative treatment currently available. To investigate resistance mechanisms to temozolomide (TMZ), the standard-of-care chemotherapy, we generated perturbed proteomic data (with and without TMZ) for 12 patient-derived cell lines (PDCLs). Pathways enrichment and independent component analysis revealed a high inter-patient heterogeneity. Commonly dysregulated proteins across PDCLs were matched to pharmacological compounds targeting them using a new pipeline. These molecules are currently being evaluated in a dedicated drug screening. The next step is to build a digital twin for each PDCL, enabling personalized prediction of combination therapies. A previously developed quantitative systems pharmacology (QSP) model of TMZ pharmacokinetics-pharmacodynamics serves as a foundation. To initiate model individualization, we developed a method to personalize model parameters using publicly available multi-omics and TMZ cytotoxicity data. Current work focuses on integrating proteomics-derived key species into the core model, via network reconstruction methods, to obtain PDCL-specific digital twins and infer personalized treatment by combining QSP and machine learning. This integrative approach, combining data analysis, network reconstruction, and mechanistic modeling, opens the path for efficient patient specific therapies in GBM.

Organized by: Wenjun Zhao (UBC/Wake Forest University), Khanh Dao Duc (UBC)

Note: this minisymposia has multiple sessions. The other session is: and Part-2.

1. Ben Cardoen University of Birmingham
   "Shape discovery of functional interaction between proteins and organelles in the presence of weak oracle distances in superresolution microscopy"
Novel superresolution microscopy (SRM) allows mesoscale (5-150 nm) discovery in situ, in both live and fixed cells. Unlike EM based approaches, it is less costly, less invasive, and enables tagging of individual targets with fluorescence, at a cost of lower precision. Multichannel SRM enables the study of interacting organelles and protein complexes with use cases such as : ER-Mitochondria contacts, HIV ingress, HIV coat forming, protein complex formation dynamics, chromatic dynamics, and neurotransmitter patterns. Interaction at mesoscale is defined as distance mediated state change. Where EM based analysis is ideally placed to reconstruct stable structure, SRM can describe equilibria and diversity. However, SRM is characterized by complex non-additive noise, and localizes objects with an uncertainty and precision that can be as high as the size of or distance between objects. In other words, SRM interaction analysis works with weak distance oracles. Second, the physics at the mesoscale are decidedly non-linear, calling for algorithms that leverage these factors. Finally, a number of underappreciated SRM specific confounding factors can disrupt downstream analysis. In this talk I will give an overview of those challenges, and review how current methods elucidate interaction patterns from SRM data. Using a new computational paradigm to formalize interaction mathematically, I will review underappreciated confounding factors that can comprise SRM interaction analysis. Finally, using in silico data I will illustrate the potential and limitations of current computational techniques to recover distance mediated state change from SRM data. We will measure if we can detect pentagonal versus hexagonal protein conformations, typical in membrane coat function to form spherical structures, or in the capsid coat of the HIV1 virus.
2. Ashok Prasad Colorado State University
   "Static Shapes and Dynamic Networks: Morphological Analysis of Cellular Identity"
Cell morphology offers a powerful and underutilized lens for understanding cellular identity, behavior, and state. While transcriptomic and proteomic profiling have revolutionized our capacity to characterize cells, quantitative morphological features can provide complementary insights into cell state and function. In this talk, I will present our recent work demonstrating that cells can be robustly classified using a range of morphological metrics derived from microscopy images. I will also discuss our ongoing efforts to develop morphological features that are sensitive to the dynamics of intracellular structures, such as the actin cytoskeleton and other dynamic polymer networks. We simulate the actin cytoskeleton, incorporating the action of molecular motors and cross-linkers, and look for features that are sensitive to different initial conditions and differences in temporal dynamics. Ultimately, we seek to build a framework in which cellular morphology is treated as a high-dimensional, information-rich signature of cell state. This work contributes to a broader vision of morphology-based phenotyping as a bridge between structure and function in living systems.
3. Felix Zhou UT Southwestern
   "Methods to identify causal links between morphology and cell signaling"
Form is function. Just as Darwin’s finches have beaks adapted to their ecological niche, so too do cell morphology associate with its function. Indeed, cell shape changes are widely used as a first clinical indicator of disease. Conventionally, we have thought of shape as downstream of a cell’s molecular processes. However, recently we have found that the shape of protrusions on cell surfaces might also directly drive signaling whereby changes to their properties, such as curvature and thickness dynamically in time, modify signaling cascades and ultimately affect fate. For example, we found a previously undescribed role of blebbing - dynamic hemispherical protrusions in melanoma cells to activate prosurvival signals and avoid the normal checkpoint program of programmed cell death – a prerequisite step for cancer metastasis. Causal investigation of shape and signaling is notoriously difficult due to the intricate feedback between the two. Notably, shape changes are a product of molecular signals. Consequently, we have been developing statistical causal inference techniques to systematically test for causal links between 3D cell shapes segmented from microscopy videos with jointly measured molecular signal intensities from fluorescent biosensors. Here, I will talk about 3 general computational frameworks we have developed to enable this: u-Segment3D to segment the 3D surface, leveraging pretrained generalist 2D segmentation models; u-Unwrap3D to bidirectionally map the segmented 3D surface to a 2D image; and u-InfoTrace to adapt 1D causal measures and test spatiotemporal causality in the 2D unwrapped images. I will demonstrate example application to diverse videos of 3D cell blebbing, cancer-immune cell interaction, and organoids.
4. Joe Kileel UT Austin
   "Method of moments for determining macromolecular shapes in cryo-EM"
In this talk, I will present method of moments based approaches for 3D reconstruction of molecular conformations from datasets of noisy 2D images in cryo-electron microscopy. I will present progress both theoretically and computationally for these methods, in particular leveraging prior and side information to improve the cryo-EM reconstruction. Method of moments based solvers also provide a more general methodology, and may be applicable to other inverse problems involving shape data.

Organized by: Jinsu Kim (POSTECH), Eric Foxall (The University of British Columbia - Okanagan Campus), and Linh Huynh (Dartmouth College)

Note: this minisymposia has multiple sessions. The other sessions are: Part-1, Part-3, and Part-4.

1. Ellen Baake Bielefeld University
   "Evolving genealogies in cultural evolution"
We consider a Moran-type model of cultural evolution, which describes how traits emerge, are transmitted, and get lost in populations. Our analysis focuses on the underlying cultural genealogies; they were first described by Aguilar and Ghirlanda (2015) and are closely related to the ancestral selection graph of population genetics, wherefore we call them emph{ancestral learning graphs}. We investigate their dynamical behaviour, that is, we are concerned with emph{evolving genealogies}. In particular, we consider the total length of the genealogy of a sample of individuals from a stationary population as a function of the (forward) time at which the sample is taken. This quantity shows a sawtooth-like dynamics with linear increase interrupted by collapses to near-zero at random times. We relate this to the metastable behaviour of the stochastic logistic model, which, in our context, describes the evolution of the number of ancestors, or equivalently, the number of descendants of a given sample. This is joint work with Joe Wakano (Tokyo), Hisashi Ohtsuki (Hayama), and Yutaka Kobayashi (Kochi).
2. Linh Huynh Dartmouth College
   "Spin glass model for Large Language Models and evolution"
In recent years, Large Language Models (LLMs) have revolutionized Natural Language Processing with their ability to generate human-like texts. However, a fundamental challenge remains in understanding the underlying mechanisms driving their emergent behaviors, particularly the randomness in their outputs. In this talk, I will discuss the application of spin glass theory as a mathematical framework to quantify the uncertainty of LLMs. By making connections between LLMs and spin glass models, which are traditionally used in statistical physics and probability to describe disordered networks with random interactions and frustrations (conflicting constraints), we can gain insights into the high-dimensional optimization landscapes of LLMs, the uncertainty in their outputs, and the role of noise in their learning process. I will conclude by making a connection to evolution.
3. Samuel Isaacson Boston University
   "Coarse-grained limits of particle-based stochastic reactive-transport models"
In many applications, both spatial transport and stochasticity in chemical reaction processes play critical roles in system dynamics. Particle-based stochastic reaction diffusion (PBSRD) models have been successfully use to study a variety of such reaction processes, particularly at the single-cell scale. However, as commonly used, they typically assume overdamped transport, ignoring inertial forces. In this talk we investigate how to construct more microscopic particle-based reactive Langevin Dynamics (PBRLD) models that include inertial forces, formulating models that are consistent with detailed balance of reaction fluxes at equilibrium. We show via asymptotic analysis that with appropriate scaling assumptions for the dependence of reaction kernels on friction/mass, PBRLD models converge to common PBSRD models in the overdamped limit. Finally, we identify and prove the large population mean-field limit of the new PBRLD models, obtaining systems of nonlocal kinetic reaction-diffusion equations.
4. Clément Soubrier University of British Columbia
   "Modeling the meiotic spindle using a spatial birth-death process."
In eukaryotes, during the second phase of meiosis, the two chromatids of each chromosomes are separated to form haploid gametes. This segregation is driven by a bi-polar mechanical and dynamical structure, the spindle, primarily composed of microtubules. Spindle defects, such as loss or split of a pole, lead to failure of the mitosis or to aneuploid gametes. In this talk, we model the spindle stability using a spatial birth-death process representing the position of micro-tubules attached to the spindle pole. In particular, we study the first transition of the process to a multipolar state. We define this state as having a large spatial gap between two consecutive micro-tubules. Our main result is an asymptotic estimate of the first passage time of the multipolar state, as a function of the spindle creation rate and spatial gap.

Organized by: Margherita Maria Ferrari (University of Manitoba), Daniel Cruz, University of Florida

1. Stacey Smith? The University of Ottawa
   "The implications of micro-host--pathogen co-evolutionary outcomes on macro-epidemics"
Host defence and pathogen virulence both interplay and mutually influence the evolutionary processes of each another. Host–pathogen co-evolutionary outcomes have potentially significant impacts on population dynamics and vice versa. To investigate host–pathogen interactions and explore the impact of micro-level co-evolutionary outcomes on macro-level epidemics, we develop a co-evolutionary model with a mixed host-defence strategy. Our results illustrate that host–pathogen co-evolution may induce infection cycling and lead to the vanishing of the disease-induced hydra effect, whereas pathogen mono-evolution strengthens the hydra effect in both range and magnitude. As the recovery rate increases, we find a counter-intuitive effect of increased disease prevalence due to host–pathogen co-evolution: the disease is first highly infectious and lethal, then highly infectious but with low lethality. Such diverse outcomes suggest that this combined co-evolutionary and epidemiological framework holds great promise for a better understanding of disease infection.
2. Morgan Craig Université de Montréal
   "Age-related variability in antibody responses to the mRNA COVID-19 vaccine primary series"
Immunological heterogeneity, driven by a variety of factors including e.g., age and sex, heavily influences vaccine outcomes. To better understand this variability, we recently developed a mechanistic mathematical model describing the generation and maintenance of humoral immunity after the mRNA COVID-19 vaccine primary series. By fitting our model to a clinical cohort of younger health care workers and seniors, we disentangled the mechanisms driving weaker antibody responses and faster antibody waning in older adults. Based on these results, we outlined vaccine strategies tailored to key characteristics driving outcomes using an approach rooted in computational immunology.
3. Chris Soteros University of Saskatchewan
   "Lattice polygon models of DNA topology"
The field of DNA Topology includes the study of DNA geometry (supercoiling) and topology (knots and links) and their effects on DNA in vitro and in vivo. Statistical physics-based lattice models of DNA have proved useful for addressing many questions arising from DNA topology experiments. In this talk I will review recent advances we have made using lattice polygon models to address questions related to the knot and link statistics of DNA in vitro either subject to varying salt conditions or under nanochannel-like confinement.
4. Margherita Maria Ferrari University of Manitoba
   "Discrete models for DNA-RNA complexes"
R-loops are three-stranded structures formed by a DNA-RNA hybrid and a single strand of DNA, often appearing during transcription. Experimental works show that R-loops can threaten genome integrity, while also playing regulatory roles in biological processes. In this talk, we introduce a model for R-loops based on formal grammars, that are systems to generate words widely applied in molecular biology. The model is trained on experimental data and, despite not including topological information, it accurately predicts R-loop formation on plasmids with varying starting topologies.

Organized by: Annequa Sundus (Indiana University Bloomington), Elmar Bucher (Indiana University Bloomington), Paul Macklin (Indiana University Bloomington)

1. Harsh Jain University of Minnesota Duluth
   "The SMoRe-verse: A novel method for ABM parametrization and uncertainty quantification"
Agent-based models (ABMs) are widely used to study complex biological systems where emergent behaviors arise from individual-level interactions. Understanding the influence of input parameters on model output is essential for interpreting results and improving predictive power, but global sensitivity analysis (GSA) remains computationally prohibitive for many ABMs due to their complexity and high simulation costs. This talk presents SMoRe GloS (Surrogate Modeling for Recapitulating Global Sensitivity), a novel, computationally efficient method for performing GSA on ABMs. SMoRe GloS leverages explicitly formulated surrogate models to approximate ABM outputs, enabling thorough exploration of parameter space and quantification of uncertainty with significantly reduced computational demands. We demonstrate the method’s flexibility and accuracy using two case studies: a 2D cell proliferation assay and a 3D vascular tumor growth model. In both settings, SMoRe GloS produced sensitivity indices consistent with established methods such as Morris one-at-a-time and eFAST, while achieving substantial reductions in computation time. Importantly, the method also captures sensitivities for parameters associated with processes not explicitly included in the surrogate model. These results highlight the potential of SMoRe GloS to extend the accessibility of GSA for computationally intensive ABMs and to support more robust model-based inference in complex systems.
2. Atul Deshpande Johns Hopkins University
   "Deriving Spatially-Resolved Descriptors from Omics Data to Inform Agent-Based Models"
This talk will highlight spatial bioinformatics methodologies that directly inform the design and enhance the calibration of Agent-Based Models (ABMs). I will introduce novel computational methods developed in our group for analyzing spatial omics data, focusing on their application in deriving spatially-resolved quantitative descriptors of tumor organization and oncogenic signaling. For instance, SpaceMarkers quantifies the spatial overlap between distinct cell types in spatial transcriptomics and identifies the molecular changes associated with these interactions. Similarly, FunCN models cellular influences on individual cells to define functional neighborhoods (CNs), allowing us to map abundant CNs in tissue samples. The descriptors from these methods are crucial not only as direct inputs for defining agent properties, interaction rules, and environmental conditions in ABMs, but also for precise model parameter calibration. I will showcase our integrated approach, applying these bioinformatics methods on spatial data from biospecimens and ABM simulations. This synergy enables advanced model calibration and validation, surpassing traditional coarse-grained techniques, and we are applying it to power virtual trials of combination therapies for hepatocellular carcinoma.

Organized by: Tomas Helikar (University of Nebraska - Lincoln), Juilee Thakar (Juilee_Thakar@URMC.Rochester.edu) - University of Rochester Medical Center James Glazier (jaglazier@gmail.com) - Indiana University

Note: this minisymposia has multiple sessions. The other sessions are: Part-1, and Part-3.

1. Reinhard Laubenbacher University of Florida
   "Immune Digital Twins: Foundational Mathematical Challenges"
The digital twin concept has its origins in industry. One industrial equipment manufacturer advertised its digital twin capabilities to its customers as ”No unplanned downtime” for its products. There is a compelling aspirational analog in healthcare: “No unplanned doctor visits.' Of course, the challenges of building digital twins for human patients are incomparably greater than for machinery. Nonetheless, there are now several instances of what might be called digital twins in medicine, and many more ongoing development projects. Aside from our incomplete understanding of human biology, relative sparseness of data characterizing human patients, and logistical difficulties in implementing computational models in healthcare, there are many mathematical and computational problems that need to be solved. Examples include calibration and validation of multiscale, hybrid, stochastic computational models, forecasting algorithms, and optimal control methods. This talk will describe some of these problems and outline a mathematical research program for the field.
2. Josh Loecker University of Nebraska-Lincoln
   "Adaptive Analysis of Mechanistic Models using Large Language Models"
Large language models (LLMs) hold immense potential for revolutionizing biomedical research and personalized medicine, but their application to mechanistic modeling and immune digital twins (IDTs) remains largely unexplored. This work proposes a novel framework integrating LLMs with mechanistic models to address two critical gaps: (1) translating complex model outputs into actionable insights for patients and clinicians, and (2) automating the analysis and interpretation of large-scale mechanistic models. Our framework leverages a comprehensive library of “Action Intents,” enabling LLMs to interact with and manipulate models, perform complex analyses, and generate human-readable explanations. We will develop novel LLM-driven algorithms for tasks such as parameter sensitivity analysis, critical node identification, and emergent behavior prediction. Furthermore, we will establish robust evaluation metrics to assess LLM performance in this domain, encompassing both quantitative measures of accuracy and qualitative assessments of clinical utility. This framework will empower patients with personalized, understandable insights derived from their Personalized Digital Twin, fostering greater autonomy in healthcare decisions. Simultaneously, it will provide researchers with powerful tools to accelerate the analysis and interpretation of complex biological models, ultimately advancing our understanding of the immune system and accelerating the development of novel therapeutic strategies. This innovative approach promises to bridge the gap between complex biological models and their practical application in personalized medicine, paving the way for more effective and patient-centered healthcare.
3. Heber L. Rocha Indiana University
   "Multiscale Modeling of Immune Surveillance for Cancer Patient Digital Twins"
Tumors are complex ecosystems characterized by heterogeneous cellular behaviors, intercellular interactions, and stochastic processes, which collectively challenge the development of personalized cancer therapies due to unpredictable therapeutic responses. Digital twins—computational representations of individual patients—offer a transformative approach to simulate and predict treatment outcomes, enabling precision oncology. This presentation describes an multiscale agent-based model, developed using the PhysiCell framework, to investigate immune surveillance in micrometastases, early metastatic clusters critical to cancer progression. Through high-throughput simulations of over 100,000 virtual patient trajectories, our model revealed a spectrum of outcomes, ranging from tumor proliferation to immune-mediated eradication. These analyses identified critical parameters, such as immune cell functionality and tumor immunogenicity, that govern these divergent dynamics. These findings provide a robust foundation for constructing cancer patient digital twins to optimize therapeutic strategies. To enhance model reliability, our ongoing efforts focus on uncertainty quantification, employing sensitivity analysis and parameter calibration to address inherent biological variability and epistemic uncertainties, thereby advancing the development of clinically actionable digital twins.
4. James A. Glazier Indiana University, Bloomington
   "Medical Digital Twins: Addressing Simulation Equivalence Challenges in Virtual-Tissue Models"
Developing closed-loop Medical Digital Twins requires multiple tools—both physical (sensors/actuators) and computational—to support the cycle of measurement, forecasting, divergence assessment, anomaly detection, data assimilation, and action selection. While significant progress has been made in predictive modeling and data assimilation, comparing simulation states presents unique challenges, particularly for agent-based spatial Virtual-Tissue models. When working with scalar quantities like blood oxygenation, comparing measured and forecast values is straightforward. However, for Virtual-Tissue models, determining whether two simulation states derive from the same underlying model becomes complex. Implementation differences across software frameworks create substantial numerical variations (inter-simulation variability), while stochasticity within single implementations produces multiple potential phenotypes (intra-simulation variability). To address these reproducibility and interoperability challenges, I present three methodologies for determining simulation state equivalence despite phenotypic differences: 1) A neural-network image classifier that learns features of equivalent model configurations robust to both intra- and inter-simulation variability. This classifier also supports developing generative AI surrogates of mechanistic agent-based models for Medical Digital Twin applications. 2) AI/ML approaches to cluster and classify synthetic images generated by agent-based models of cell sorting and angiogenesis. 3) Leveraging the classification techniques to solve the inverse problem of inferring model parameters from images, enabling parameter identification in complex systems. The presentation concludes with proposed next steps for advancing these techniques in the Digital Twin ecosystem.

Organized by: Hye-Won Kang (University of Maryland Baltimore County), Arnab Ganguly, Louisiana State University, aganguly@lsu.edu

Note: this minisymposia has multiple sessions. The other sessions are: Part-1, and Part-3.

1. Joshua McGinnis University of Pennsylvania
   "Homogenization of a Spatially Extended, Stochastic Ion Channel Model"
Simulations of stochastic neuron potential models, which describe the voltage potential along the length of a neuron’s axon and incorporate ion channel noise as Gaussian fluctuations, have shown that channel noise can induce complex phenomena such as jitters and splitting of action potentials [1] and place constraints on the miniaturization of axons [2]. To develop a robust analytic framework for understanding stochastic effects of channel noise on action potential propagation in a neuron, we need to begin by investigating how many independent, spatially distributed ion channels can collectively yield deterministic behavior. We start with an electrophysiological derivation of a simple discrete model and contrast this with a common, yet less physically accurate approach where the law of large numbers and the central limit theorem are more easily applied. Our model couples a spatially discretized diffusive PDE for the voltage with continuous-time Markov processes that govern the behavior of the ion channels. We will then outline an argument using homogenization theory to estimate the rate of strong convergence to the typical deterministic PDE as the spacing between ion channels approaches zero. Finally, we present a numerical technique for simulating our model and discuss the challenges involved in increasing computational efficiency of simulations. [1] Faisal AA, Laughlin SB. Stochastic simulations on the reliability of action potential propagation in thin axons. PLoS Comput Biol. 2007 May;3(5):e79. doi: 10.1371/journal.pcbi.0030079. PMID: 17480115; PMCID: PMC1864994. [2] Faisal AA, White JA, Laughlin SB. Ion-channel noise places limits on the miniaturization of the brain's wiring. Curr Biol. 2005 Jun 21;15(12):1143-9. doi: 10.1016/j.cub.2005.05.056. PMID: 15964281.
2. Radek Erban University of Oxford
   "Chemical Reaction Networks: Systematic Design, Limit Cycles and Spatio-Temporal Modelling"
I will discuss mathematical methods for describing biochemical reaction networks, with applications to modelling of intracellular processes. Several types of mathematical models of chemical reaction systems will be considered, including (i) deterministic models which are written in terms of reaction rate equations (i.e. ordinary differential equations (ODEs) for concentrations of chemical species involved); (ii) stochastic models of reaction networks, given in terms of the Gillespie stochastic simulation algorithm, which provides more detailed information about the simulated system than ODEs; and (iii) spatio-temporal models described by the reaction-diffusion master equation and Brownian dynamics simulations. I will discuss methods for systematic design of relatively simple reaction systems with prescribed dynamical behaviour, including reaction systems with multiple oscillating solutions (limit cycles). I will also present methods for efficient spatio-temporal modelling of intracellular processes.
3. David Lipshutz Baylor College of Medicine
   "Methods for Comparing Sensitivities of Stochastic Neural Networks"
Biological neural networks (and some artificial neural networks) transform stimuli into stochastic neural responses. Each network induces a Riemannian geometry in stimulus space via the Fisher-Rao metric and it is of interest in various applications to compare the local geometries on stimulus space that are induced by different networks. Such comparisons can be challenging when stimulus space is high-dimensional (e.g., images), so one approach is to identify a few directions in stimulus space along which to compare the induced geometries. We propose a method for optimally selecting directions in stimulus space that maximally differentiate a set of networks in terms of the induced local geometries along these directions. We apply our method to compare a set of simple models of the early visual system and show that our method produces image distortions that allow for immediate visual comparison of these models. This is joint work with Jenelle Feather, Sarah Harvey, Alex Williams and Eero Simoncelli.
4. Minji Lee Ulsan National Institute of Science and Technology
   "MPUGAT: A Novel Framework for Inferring Dynamic Infectious Disease Transmission with Graph Attention"
Estimating spatiotemporal transmission patterns in infectious disease modeling often involves high-dimensional parameters and hidden contact structures, leading to challenges in parameter identifiability. We present MPUGAT, a hybrid framework that integrates a multi-patch SEIQ compartment model with a spatio-temporal deep learning architecture utilizing Graph Attention Networks (GATs). MPUGAT dynamically estimates the time-varying transmission matrix by inferring latent interaction patterns from diverse time series data, including mobility indicators and case counts. Our approach captures stochastic variations in contact structures and transmission intensity by treating epidemic parameters as latent variables inferred under observational uncertainty. Through a case study on COVID-19 in South Korea, MPUGAT demonstrates strong alignment with real-world social distancing interventions and regional heterogeneity in disease spread. This framework illustrates how non-epidemic time series data, integrated through deep learning, can enhance the identifiability of high-dimensional transmission parameters in epidemic models.

Organized by: Jinsu Kim (POSTECH), Eric Foxall (The University of British Columbia - Okanagan Campus), and Linh Huynh (Dartmouth College)

Note: this minisymposia has multiple sessions. The other sessions are: Part-1, Part-2, and Part-4.

1. Hwai-Ray Tung University of Utah
   "Extreme first passage times with fast immigration"
Many scientific questions can be framed as asking for a first passage time (FPT), which generically describes the time it takes a random 'searcher' to find a 'target.' The important timescale in a variety of biophysical systems is the time it takes the fastest searcher(s) to find a target out of many searchers. Previous work on such fastest FPTs assumes that all searchers are initially present in the domain, which makes the problem amenable to extreme value theory. In this paper, we consider an alternative model in which searchers progressively enter the domain at some 'immigration' rate, which may be constant, time inhomogeneous, or proportional to the population size. In the fast immigration rate limit, we determine the probability distribution and moments of the k-th fastest FPT. Our rigorous theory applies to many models of stochastic motion, including random walks on discrete networks and diffusion on continuous state spaces. Mathematically, our analysis involves studying the extrema of an infinite sequence of random variables which are both not independent and not identically distributed. Our results constitute a rare instance in which extreme value statistics can be determined exactly for strongly correlated random variables.
2. Minjun Kim POSTECH
   "A Path Method for Non-exponential Ergodicity of Stochastic Chemical Reaction Systems"
We present criteria for non-exponential ergodicity of continuous-time Markov chains on a countable state space in total variation norm. These criteria can be verified by examining the ratio of transition rates over certain paths. We applied this path method to explore the non-exponential convergence of microscopic biochemical interacting systems. Using reaction network descriptions, we identified special architectures of biochemical systems for non-exponential ergodicity. In essence, we found that reactions forming a cycle in the reaction network can induce non-exponential ergodicity when they significantly dominate other reactions across infinitely many regions of the state space. Interestingly, the special architectures allowed us to construct many detailed balanced and complex balanced biochemical systems that are nonexponentially ergodic. Some of these models are low-dimensional bimolecular systems with few reactions. Thus we suggest the possibility of discovering or synthesizing stochastic systems arising in biochemistry that possess either detailed balancing or complex balancing and slowly converge to their stationary distribution.
3. Daniel Coombs University of British Columbia
   "Resolving Single Particle Mobility with Hidden Markov Models"
Using single particle tracking approaches, we study the motion of individual molecules on cell surfaces. At larger space- and time-scales, we can also follow the motion of individual cells in vitro and in vivo. In this talk I will describe how we have applied hidden Markov models to provide a time-resolved view of particle behaviour. In this talk I will describe how our approach has developed from the simplest two-state model, to inferring multi-state models and incorporating noise using an infinite hidden Markov model framework.

Organized by: Tomas Helikar (University of Nebraska - Lincoln), Juilee Thakar (Juilee_Thakar@URMC.Rochester.edu) - University of Rochester Medical Center James Glazier (jaglazier@gmail.com) - Indiana University

Note: this minisymposia has multiple sessions. The other sessions are: Part-1, and Part-2.

1. Juilee Thakar University of Rochester
   "Monocyte digital twin and HIV associated vascular disease"
People living with HIV (PLWH) continue to show a heightened risk for atherosclerosis (AS) even under effective antiretroviral therapy (ART). Monocytes are key drivers of AS pathogenesis. They can directly contribute to lesion formation by differentiating into lipid-laden macrophages (foam cells) in the arterial intima. Indirectly, their persistent immune activation and secretion of inflammatory cytokines support chronic inflammation, a hallmark of HIV-associated vascular disease. Because monocytes continuously replenish the macrophage pool in the vessel wall, they represent an important early predictor of AS progression in HIV. To investigate this, we performed single-cell transcriptomic profiling of 138,487 circulating monocytes from four well-matched participant groups—HIV-AS−, HIV-AS+, HIV+AS−, and HIV+AS+—stratified by age, sex, and Reynolds cardiovascular risk score. We identified eight transcriptionally distinct monocyte subsets, including canonical CD14+ cells and a previously undescribed population characterized by platelet interaction, referred to as platelet-monocyte complexes (PMCs). We used Boolean Omics Network Invariant Time Analysis (BONITA) developed in our group to identify pathway specific stable cellular states and their basin of attraction. Using these cellular states we have defined monocyte digital twins that predict the AS pathogenesis.
2. Esteban Hernandez Vargas University of Idaho
   "Adaptive Observers in Digital Twins for Drug Resistance Mitigation in HIV"
High mutation rates in HIV pose a significant challenge for long-term therapy, as the virus can quickly develop resistance to specific antiretroviral drugs. Despite extensive research, there remains no clear consensus on how to schedule treatments to maintain viral suppression and mitigate resistance optimally. In this talk, I present a digital twin framework for modeling HIV mutation dynamics, employing an adaptive observer to approximate a surrogate of a higher-order nonlinear mutation model. This approach enables us to monitor and anticipate the emergence of drug-resistant strains in silico, providing a foundation for exploring adaptive treatment strategies. Preliminary simulation results indicate that this computational framework can outperform standard clinical scheduling recommendations, offering a more individualized and responsive alternative to conventional therapy. This work represents a step toward leveraging digital twins to support clinical decision-making in the treatment of complex, mutating viral infections. Funding: This research was supported by the National Science Foundation grant DMS -2315862.
3. Elsje Pienaar Purdue University
   "Patient-specific Immuno-profiles in Mechanistic Models: CD8+ T cell Exhaustion in children with perinatal HIV"
We and others have reported evidence of T cell exhaustion in children with perinatal HIV with increased expression of inhibitory receptors PD-1, CD160, and TIM-3, but there is limited data on the virologic functional consequences of this immune exhaustion. We address this by using an immune database from Kenyan children with perinatal HIV and unexposed controls. We computationally integrate T cell profiles of differentiation, activation and exhaustion in an agent-based model (ABM) to predict how T cell exhaustion impacts viral control following HIV exposure in vitro. Our ABM includes macrophages, CD4 and CD8 T cells, cytokines, and HIV. Model mechanisms include viral dynamics, macrophage activation, T cell activation and proliferation, cytotoxic T cell killing, and cytokine/HIV diffusion and degradation. Participants are grouped by HIV plasma viremia and by age, less than 5 years or 5-18 years. Our findings indicate that cells from virally active participants, who have the highest levels of exhaustion, have lower predicted viral concentrations and infected cells compared to other participant groups during new infection. However, this coincides with higher cell death, suggesting that short-term viral control is associated with excessive inflammation, which could be detrimental long-term. Cells from virally suppressed participants older than 5 years can maintain lower viral concentrations while limiting cell death, reflecting a more sustainable short-term immune response. In virally suppressed children younger than 5 years, immune response patterns strongly resemble the age-matched healthy control group, suggesting early viral suppression may preserve antiviral immune responses. Our model predicts unique patterns of cell death for each participant group, with CD8 T cell death being dominant in virally active groups and CD4 T cell and macrophage death being dominant in healthy and virally suppressed groups. Finally, exhausted CD8 T cells are predicted to contribute significantly to CD8 T cell killing, proliferation, and activation in the virally active group, indicating partially functional CD8 T cells can still contribute to short-term viral control. Our analysis functionally integrates participant-specific immunophenotypic data to allow quantification of the extent, mechanisms, and impact of immune dysfunction in perinatal HIV and could inform pediatric HIV remission and cure strategies.
4. Miriam Rafailovich
   "Molecular level modeling of the immune response and thrombosis following viral infection"
Abstract to be determined. Please check back later.

Organized by: Hye-Won Kang (University of Maryland Baltimore County), Arnab Ganguly, Louisiana State University, aganguly@lsu.edu

Note: this minisymposia has multiple sessions. The other sessions are: Part-1, and Part-2.

1. Suzanne Sindi University of California Merced
   "Scalable Bayesian Discovery of Chemical Reaction Networks from Fully Observed Stochastic Dynamics"
We present a Bayesian framework for inferring chemical reaction networks (CRNs) from fully observed state-transition data, using spike-and-slab priors to jointly model reaction rates and network structure with uncertainty-aware sparsity. Building on previous work, we leverage likelihood decomposition to enable scalable inference, and demonstrate practical identifiability in three-species networks where the full reaction set is combinatorially large. Our method captures nontrivial posterior structure even in intermediate data regimes, where traditional MLE-based sparse regression methods may fail due to over-penalization of small but important reactions. Unlike point-estimate approaches, our model returns full posterior distributions, allowing principled model selection. We show that our approach generalizes to higher-dimensional systems by exploiting structural sparsity and decomposition strategies, providing a tractable path toward Bayesian inference in large, complex reaction networks.
2. Muruhan Rathinam University of Maryland Baltimore County
   "Stochastic Filtering of Reaction Networks"
We consider the problem of inferring states and/or parameters from exact observations of a subset of states of a stochastic reaction network. We present two particle filtering methods for the computation of the conditional distribution of the state and/or parameters, one for the case of continuous in time observations and the other for the case of observations in discrete snapshots of time. In addition to presenting theoretical justification, we also provide numerical examples to illustrate the applicability of these methods.
3. Arnab Ganguly Louisiana State University
   "Multiscale Enzyme Kinetic Reactions: Stochastic Averaging and Statistical Inference"
We study a stochastic model of multistage enzyme kinetics of the Michaelis–Menten (MM) type, where substrate molecules are converted into product through a sequence of intermediate species. The reaction network is both high-dimensional and multiscale, posing substantial computational challenges, particularly in estimating reaction rates. These challenges are compounded when direct observations of the system's states are unavailable and only random samples of product formation times are accessible. To address this, we adopt a two-stage approach. In the first stage, under technical assumptions similar to those in the Quasi-Steady-State Approximation (QSSA) literature, we establish two asymptotic results: a stochastic averaging principle that reduces the model’s dimensionality, and a functional central limit theorem that characterizes the resulting fluctuations. In the second stage, we consider the problem of estimating parameters of the system from data consisting of a sample of product-formation times. Note that such a dataset does not allow reconstruction of temporal paths of species rendering any trajectory- based inference method categorically inapplicable. To address this, we develop a novel inference framework based on an interacting particle system (IPS) that approximately captures the dynamics the reduced-order model at a molecular level. The crux of our approach is a propagation of chaos result that leads to an asymptotically exact product-form expression for the likelihood function. Numerical examples are presented to demonstrate the effectiveness of the proposed approach. This is a joint work with Wasiur R. KhudaBukhsh.
4. Boseung Choi Korea University Sejong Campus
   "Statistical Inference Method for Identifying the Stochastic Chemical Kinetics Using Logistic Regression"
Identifying network structures and inferring parameters are challenging tasks in the modeling of chemical reaction networks. This study presents likelihood-based methods that utilize logistic regression to derive these components from complete time-series data of stochastic chemical reaction networks. When full trajectories of molecular counts for all species are available, the stoichiometries can be identified, provided that each reaction occurs at least once during the observation period. However, determining which species act as catalysts is more difficult since their molecular counts do not change with the occurrence of reactions. We demonstrate the effectiveness of logistic regression in identifying the entire network structure, including stoichiometric information, using three stochastic models that incorporate catalytic reactions. Additionally, we investigate Bayesian logistic regression approaches for estimating model parameters using real epidemic data. To tackle the challenges presented by data observed from only a subset of populations, we propose a method that combines Bayesian logistic regression with differential equations to infer parameters in the SIR model, utilizing COVID-19 infection data. Our findings emphasize the potential of straightforward likelihood-based methods, such as logistic regression, to extract valuable modeling insights from both synthetic and real-time series data.

Organized by: Jinsu Kim (POSTECH), Eric Foxall (The University of British Columbia - Okanagan Campus), and Linh Huynh (Dartmouth College)

Note: this minisymposia has multiple sessions. The other sessions are: Part-1, Part-2, and Part-3.

Note: this minisymposia has been accepted, but the abstracts have not yet been finalized.

Organized by: Cody FitzGerald (Northwestern University), Rainey Lyons (CU Boulder), Nora Heitzman-Breen (CU Boulder), Susan Rogowski (NCSU)

Note: this minisymposia has multiple sessions. The other sessions are: Part-1, Part-2, and Part-3.

Note: this minisymposia has been accepted, but the abstracts have not yet been finalized.

Organized by: James Osborne (University of Melbourne), James Glazier (Indiana University) Yi Jiang (Georgia State University)

Note: this minisymposia has multiple sessions. The other sessions are: Part-1, and Part-2.

1. James Glazier Indiana University, USA
   "OpenVT--Towards Making Virtual Tissue Models FAIR - Opportunities and Challenges"
Multi-scale, Multicellular Agent-Based Virtual-Tissue models built using modeling frameworks like CompuCell3D, Morpheus, Artistoo, CHASTE or PhysiCell are versatile tools for exploring the complex interactions between intracellular signaling and gene-regulatory networks, inter-cellular signaling through contact and diffusible signals, and force generation, cell migration and shape change. They can play a crucial role in helping to interpret and design more informative experiments, in particular in in vitro to in vivo extrapolation. However, Virtual Tissues currently lack the model-specification standards, support for modular architectures and annotation, cross-compatible tools for graphical model specification, visualization and analysis and accepted model sharing infrastructure that have enabled the rapid developing of systems biology network modeling as a core technology in modern biology and the regulatory acceptance of these approaches. Comparable infrastructure is essential for Virtual Tissues to move from academic one-offs for discovery science to truly progressive mainstream technologies in biomedicine, technology and regulation. Because Virtual Tissues are substantially more complex and structurally and functionally diverse than network models, standardization and modularization, graphical specification and distribution are all more challenging. I will consider some of the variety of Virtual Tissue applications, frameworks and modeling approaches and some of the challenges and opportunities we face in developing an effective ecosystem of tools and standards. I will also discuss how the NSF-funded OpenVT project is working to build community to address these challenges.
2. Paul Macklin Indiana University, USA
   "Intuitive code-free tissue modeling in the cloud with PhysiCell"
Agent-based models (ABMs) simulate individual cells as they move and interact in a virtualized tissue microenvironment (TME). When developing an ABM for a complex multicellular system, a scientist must define diffusible chemical substrates (e.g., oxygen and signaling factors), cell types, and functional relationships between cell behaviors and the chemical and physical signals in the simulated tissue environment. To date, creating an ABM requires scientists to encode these relationships–the “rules” of the cell agents–by hand: first as qualitative statements, then as mathematics, and finally as custom-written simulation code. As a result, ABMs take substantial time to develop and debug, and their code is neither interpretable nor reusable. In this talk, we describe a new (recently published), intuitive cell behavior grammar that writes ABM rules with human-interpretable language (e.g., “IL6 increases migration speed”), and directly and uniquely transforms these interpretable statements into mathematics and model code at run-time without need for hand coding. We also show a graphical studio (PhysiCell Studio) that allows scientist users to rapidly create, explore, and refine these code-free models on the desktop or in the cloud. We show examples from cancer hypoxia, immunology, neurodevelopment, and combination cancer treatments. Beyond the reference implementation in the PhysiCell ABM framework, the modeling grammar could provide a basis for model annotation and exchange between open source simulation toolkits, including “virtual cell templates” (digital cell lines) that bundle a cell type with base behavioral parameter values and cell rules written in this grammar.
3. James Osborne University of Melbourne, AUSTRALIA
   "Multicellular Model Reproducibility: A case study, results from the Open VT hackathon"
Multicellular development is a key area of ongoing research, focussing on how tissues and organs develop and function, particularly how underlying processes fail. The last decade has seen remarkable progress in experimental studies of tissue and organ development, leading to the development of more advanced mathematical models and increased computational power. This has enabled the increased adoption of multicellular approaches to modelling the self-organisation of cells within tissues. Multicellular simulations have become indispensable in understanding complex biological phenomena, from tissue development to disease progression. However, the diversity in simulation methods, such as agent-based models, lattice-free models, and stochastic particle simulations, poses challenges in terms of reproducibility, modularity, reusability, and integration within multi-scale multicellular simulations. To address these challenges, we organised a workshop titled “Community Development of Multicellular Virtual Tissues: The OpenVT project” on the 13th of July 2025, as part of the 2025 SMB meeting. We held a Reproducibility Hackathon during the workshop to test model reproducibility and specification. In this talk, we provide an overview of the workshop and present our findings on reproducibility in multicellular simulation.

Organized by: Jennifer Flegg (University of Melbourne), Prof Mat Simpson, Queensland University of Technology

Note: this minisymposia has multiple sessions. The other session is: and Part-1.

1. Zan Luthey-Schulten University of Illinois at Urbana-Champaign
   "Bringing a cell to life on a computer and in Minecraft"
I will describe our research into constructing 4D (x,y,z + time) models of a living minimal cell. The 4D simulations integrate data from -omics, cryo-electron tomograms, DNA contact maps, fluorescent imaging, and kinetic experiments to initialize a realistic cell state as well as validate the states as they progress in time. Fundamental behaviors emerge from these simulations that reveal how the cell balances the demands of its metabolism, genetic information processes, and growth, oQering insight into the principles of life. Validation by coarse-grained atomistic MD simulations and experiments are critical steps in building func2oning models for bacterial and eukaryo2c cells. As part of the education and knowledge transfer goals of the NSF STC for Quantitative Cell Biology, we are bringing these simulations to Minecraft, enabling players to explore a whole living cell in an immersive 3D environment.
2. Hilary Hunt Queensland University of Technology
   "Stress, stability, and systems biology: Modelling yeast’s mRNA panic rooms"
Messenger RNA (mRNA) is the biochemical link between genetic information and protein synthesis. Experimentally measuring the amount of mRNA present in the cell for each gene (transcriptomics) has become relatively cheap and reliable, especially compared to measurements of downstream processes like protein abundance or enzyme activity. However, the mapping from the amount of mRNA present in a cell to the amount of protein produced is inconsistent between mRNA species. Between mRNA transcription from DNA and its subsequent translation into protein, there are multiple regulatory processes that affect each molecule’s lifespan and rate of translation. We are particularly interested in how stress granules affect mRNA survival. Under specific environmental conditions, mRNA can be sequestered into phase-separated compartments known as stress granules and physically removed from other regulatory mechanisms. Using a minimal model of mRNA dynamics and post-transcriptional modifications, we explore the effect these granules have on mRNA distributions in yeast, factors that impact which molecules are protected when the cell is under pressure, and how this might improve our transcriptome to proteome mappings.
3. Abigail Kushnir University of Edinburgh
   "Effective Mesoscopic Rate Equations for Spatial Stochastic Systems"
Chemical master equations (CMEs) describe stochastic reaction kinetics at the mesoscopic level. Generally, their predictions for the mean molecule numbers do not agree with the predictions of the (macroscopic) deterministic rate equations. Effective mesoscopic rate equations (EMREs), derived from van Kampen's system size expansion of CMEs, correct the deterministic rate equations. Here I discuss work to extend EMREs to the spatial domain – resulting in reaction-diffusion – and discuss their implementation in the Julia programming language. I demonstrate that these spatial EMREs offer a rapid way to identify regions of parameter space where there are significant disagreements between deterministic and stochastic formulations of reaction-diffusion systems.
4. Mica Yang Stanford University
   "Whole-cell modeling of E. coli colonies enables quantification of single-cell heterogeneity in antibiotic responses"
Antibiotic response in bacterial colonies is often characterized by phenotypic heterogeneity. This heterogeneity may in turn be driven by stochastic expression of antibiotic resistance genes, linking variation in molecular-scale gene expression to population-scale phenotypes. To better understand heterogeneous antibiotic responses, we bridged the molecular and colony-level scales by embedding instances of an E. coli whole-cell model in a dynamic spatial environment model. The resulting simulations enabled us to study variations in colony-level response to two beta-lactam antibiotics with differing mechanisms of action, tetracycline and ampicillin.

Organized by: Wenjun Zhao (UBC/Wake Forest University), Khanh Dao Duc (UBC)

Note: this minisymposia has multiple sessions. The other session is: and Part-1.

1. Laurent Younes JHU
   "Aligning measures using large deformation diffeomorphic mapping for spatial transcriptomics"
Signed measures provide a powerful and flexible object representation for registration problems as they cover both continuous and singular objects and are naturally transformed by diffeomorphisms. They can, in particular, be used as tools describing functions taking values on arbitrary feature space, which makes them well adapted to the representation of spatial transcriptomic images. In this presentation, we will summarize the theoretical foundations of measure registration using large deformation diffeomorphic measure mapping, and provide applications to spatial transcriptomics, within and across modalities.
2. Luis F Pereira UCSB
   "Statistical shape analysis with Geomstats"
Geomstats is an open-source Python package for computations and statistics on Riemannian manifolds. It provides object-oriented and extensively unit-tested implementations. Manifolds can be equipped with Riemannian metrics with associated exponential and logarithmic maps, geodesics, and parallel transport. Building on this general framework, the shape module implements widely used shape spaces, such as the Kendall shape space and elastic spaces of discrete curves and surfaces, by leveraging the abstract mathematical structures of group actions, fiber bundles, and quotient spaces. The Riemannian geometry tools enable users to compare, average, and interpolate between shapes belonging to a given shape space. These essential operations can then be used to perform statistics on shape data. In this talk, we will present the object-oriented implementation of the shape module along with illustrative examples and demonstrate its use in performing statistics on shape spaces.
3. Qiyu Wang UBC
   "Studying SARS-CoV2 spike protein heterogeneity from large Cryo-EM dataset with linear subspace method and path analysis"
Recent advances in single particle cryogenic electron microscopy (cryo-EM) have allowed to capture biomolecules in various conformations through large image datasets. However, interpreting and quantifying such conformational heterogeneity remain computationally challenging, leading to a variety of recent methods. In the context of SARS-CoV-2, we developed and implemented a pipeline to process large datasets (~ millions) of 2D images of spike proteins, and apply REgularized COVARiance estimator (RECOVAR), to project the images into a latent linear subspace. Our pipeline also includes new methods for trajectory inference and transport-based segmentation that facilitate data analysis, revealing specific transitions between multiple conformations of the receptor binding domains (RBDs) in SARS-Cov2 spike protein. Our study notably led us to discover a state with three RBDs up, as well as finding a cooperativity mechanism from states with one RBD up, that goes towards the closed state before transiting to the state with two RBD’s up, offering valuable insights into the conformational landscape of SARS-CoV-2.
4. Willem Diepeveen UCLA
   "Curvature corrected tangent space-based approximation of manifold-valued data and applications in protein dynamics analysis"
When generalizing schemes for real-valued data approximation or decomposition to data living in Riemannian manifolds (widely used for modelling biological shapes), tangent space-based schemes are very attractive for the simple reason that these spaces are linear. An open challenge is to do this in such a way that the generalized scheme is applicable to general Riemannian manifolds, is global-geometry aware and is computationally feasible. Existing schemes have been unable to account for all three of these key factors at the same time. In this work, we take a systematic approach to developing a framework that is able to account for all three factors. In addition, we consider applications of our theory to analysis of protein dynamics data.

Organized by: Dhananjay Bhaskar (Yale University), Bernadette Stolz-Pretzer

Note: this minisymposia has multiple sessions. The other session is: and Part-1.

1. Eunbi Park Georgia Institute of Technology
   "Topological data analysis of pattern formation of human induced pluripotent stem cell colonies"
Understanding the multicellular organization of stem cells is vital for determining the mechanisms that coordinate cell fate decision-making during differentiation; these mechanisms range from neighbor-to-neighbor communication to tissue-level biochemical gradients. Current methods for quantifying multicellular patterning tend to capture the spatial properties of cell colonies at a fixed scale and typically rely on human annotation. We present a computational pipeline that utilizes topological data analysis to generate quantitative, multiscale descriptors which capture the shape of data extracted from 2D multichannel microscopy images. By applying our pipeline to certain stem cell colonies, we detected subtle differences in patterning that reflect distinct spatial organization associated with loss of pluripotency. These results yield insight into putative directed cellular organization and morphogen-mediated, neighbor-to-neighbor signaling. Because of its broad applicability to immunofluorescence microscopy images, our pipeline is well-positioned to serve as a general-purpose tool for the quantitative study of multicellular pattern formation.
2. Jian Tang Mila - Quebec AI Institute
   "Geometric Deep Learning for Protein Design"
Proteins are workhorses of living cells. Understanding the functions of proteins is critical to many applications such as biomedicine and synthetic biology. Thanks to recent biotechnology breakthroughs such as gene sequencing and Cryo-EM, a large amount of protein data (such as protein sequences and structures) are generated, providing a huge opportunity for AI. As the functions of proteins are determined by their structures, in this talk, I will introduce our recent work on protein understanding based on protein 3D structures with geometric deep learning. I will introduce three different topics including protein representation learning, generative models for protein structure prediction, and generative models for protein design, and also how these techniques are used for real-world applications in protein design.

1. Arianna Ceccarelli University of Oxford
   "A Bayesian inference framework to calibrate one-dimensional velocity-jump models for single-agent motion using discrete-time noisy data"
Advances in experimental techniques allow the collection of high-resolution spatio-temporal data that track individual motile entities over time and could be used to calibrate mathematical models of individual motility. However, experimental data is intrinsically discrete and noisy, and these characteristics complicate the effective calibration of models for individual motion. We consider individuals whose movement can be described by velocity-jump models in one spatial dimension, characterised by successive Markovian transitions between a network of n states, each with a specified velocity and a fixed rate of switching to every other state. We develop a Bayesian framework to calibrate these models to discrete and noisy data, which uses a likelihood consisting of approximations to the model solutions which we previously obtained. We apply the framework to recover the model parameters of simulated data, including the probabilities of switching to every other state. Moreover, we test the ability of the framework to select the most appropriate model to fit the data, including comparisons varying the number of states n.

1. James Holehouse The Santa Fe Institute
   "The Origins of Transient Bimodality"
Deterministic and stochastic models, though often used to describe the same biological systems, can yield qualitatively different predictions. In particular, deterministic bistability does not necessarily imply stochastic bimodality, and vice versa. Multistability and multimodality are typically seen as indicators of distinct system behaviors, often inferred from stochastic simulation trajectories. In this talk, I explore the disconnect between probability modes and behavioral modes in the context of transient bimodality—also known as “adiabatic explosions”—which refers to metastable probability modes that do not correspond to distinct dynamical behaviors at the trajectory level. I present recent findings that link the emergence of these transient modes to a breakdown of the central limit theorem, specifically in the context of first-passage time distributions to absorbing states. I conclude by discussing how transient bimodality challenges conventional interpretations of system behavior in biological contexts and highlight conditions under which this phenomenon becomes particularly relevant.
2. Anthony Pasion Queen's University
   "Long-Lasting and Slowly Varying Transient Dynamics in Discrete-Time Systems"
Mathematical models of ecological and epidemiological systems often focus on asymptotic dynamics, such as equilibria and periodic orbits. However, many systems exhibit long transient behaviors where certain variables of interest remain in a slowly evolving state for an extended period before undergoing rapid change. These transient dynamics can have significant implications for population persistence, disease outbreaks, and ecosystem stability. In this work, we analyze long-lasting and slowly varying transient dynamics in discrete-time systems. We extend previous theoretical frameworks by identifying conditions under which an observable of the system can exhibit prolonged transients and derive criteria for characterizing these dynamics. Our results show that specific points in the state space, analogous to transient centers in continuous-time systems, can generate and sustain long transients in discrete-time models. We further demonstrate how these properties manifest in predator-prey models and epidemiological systems, particularly in contexts where populations or disease prevalence remain low for an extended period before experiencing a sudden shift. These findings provide a foundation for understanding and predicting long transients in discrete-time ecological and epidemiological models. (Joint Work with FMG Magpantay)
3. Elmar Bucher Indiana University / Intelligent Systems Engineering
   "PhysiGym : bridging the gap between the Gymnasium reinforcement learning application interface and the PhysiCell agent-based modeling framework"
Reinforcement learning (RL) is a powerful machine learning paradigm in which an RL agent learns to discover optimal strategies in uncertain environments. The RL control strategy has achieved remarkable success in complex tasks such as playing Chess, Go, and StarCraft. For RL, the prevailing application interface (API) standard is Gymnasium, a Python library [1]. Agent-based (AB) modeling is a mathematical, dynamical system modeling approach where the parts of the system, the so-called agents, autonomously act according to agent-type specific rules. PhysiCell is an AB modeling framework written in C++ and was implemented to model multicellular systems based on Newtonian physics. Cells are the agents. The cell type specifies the rule set the agents apply. Tissue structure emerges from the cell interactions. Substrates like oxygen can be modeled with the integrated BioFVM diffusive transport solver. Additionally, intracellular models can be integrated into cell agents [2]. The resulting AB models are 2 or 3-dimensional, off-lattice, center-based, and multiscale in space and time. In this talk, we will introduce PhysiGym, a well-documented and on all major operating systems tested open-source framework written in C++ and Python that allows to control PhysiCell models over the Gymnasium API. After a brief introduction to AB models and RL, we will discuss the implementation and obtained results from our tumor microenvironment model and the RL algorithms we applied to the model. In the future, PhysiGym can be used to learn from simulations possible mechanisms that might explain how biology systems react to similar real-world control. Furthermore, if cancer patient digital twins are written as PhysiCell models, PhysiGym could ultimately be used by oncologists to explore RL reward functions to improve treatment efficacy, reduce side effects, and slow or prevent resistance. References: [1] https://gymnasium.farama.org/ , [2] https://PhysiCell.org

1. Silvia Berra IRCCS Ospedale Policlinico San Martino, Genova, Italy
   "In-silico modeling of simple enzyme kinetics: from Michaelis-Menten to microscopic rate constants"
Chemical Reaction Networks (CRNs) provide a powerful framework for modeling interactions between multiple chemical species within complex biological pathways. Their dynamics can be described by the mass-action law, representing variations over time in species concentrations through large systems of ODEs. The study of CRNs modeling signaling mechanisms within single cells has been successfully applied to provide insight into oncogenic pathways, enabling more predictive cancer models and improved therapeutic strategies [1,2,3]. Developing a CRN requires selecting the involved species, defining their interactions, and identifying key parameters such as microscopic reaction rates. This talk presents a method to retrieve these rates, particularly in the context of enzyme kinetics. We consider a simple model, where an enzyme E binds to a substrate S, forming a complex C that dissociates into a product P while regenerating E. This process is governed by four ODEs with three reaction rates: forward k_f, reverse k_r, and catalytic k_cat, typically hard to determine experimentally. A common simplification leads to the Michaelis-Menten (MM) model, where enzyme kinetics is characterized by a single ODE depending on two measurable parameters: the Michaelis constant K_M, and the maximum reaction rate V_max. These parameters may be expressed as functions of microscopic rates and are more accessible experimentally. This talk addresses the inverse problem of estimating k_f and k_r from K_M and V_max. A computational algorithm for solving this problem is presented and analyzed, along with an estimate of the reconstruction accuracy, and numerical simulations demonstrating its potential for refining kinetic models in biological and biomedical research. [1] Sommariva et al., J. Math. Biol., 82 (6): 55, 2021. [2] Berra et al., J. Optim. Theory Appl., 200 (1): 404-427, 2024. [3] Sommariva et al., Front. Syst. Biol., 3: 1207898, 2023.
2. Ismaila Muhammed Khalifa University
   "Data-driven Construction of Reduced Size Models Using Computational Singular Perturbation Method."
Most biological systems have underlying multiple spatial or temporal scales that require reduced-order models to capture their essential dynamics and analyze them. However, traditional model reduction techniques, such as Computational Singular Perturbation (CSP), rely on the availability of the governing or dynamical equations, which are often unknown from data in biomedical applications. To address this limitation, we propose a data-driven CSP framework that integrates Sparse Identification of Nonlinear Dynamics (SINDy) and Neural Networks to extract time-scale separated models directly from data. Our approach is validated on the Michaelis-Menten enzyme kinetics model, a well-established multiscale system, by identifying reduced models for the standard Quasi-Steady-State Approximation (sQSSA) and reverse Quasi-Steady-State Approximation (rQSSA). When the full model cannot be identified by SINDy due to noise, we use Neural Networks to estimate the Jacobian matrix, allowing CSP to determine the regions where reduced models are valid. We further analyze Partial Equilibrium Approximation (PEA) case, where the dynamics span both sQSSA and rQSSA regimes, requiring dataset splitting to accurately identify region-specific models. The results demonstrate that SINDy struggles in the presence of noise to identify full model from data that have underlying timescale evolution, but remains effective for identifying reduced models when dataset are partitioned correctly.
3. John Vastola Harvard University
   "Bayesian inference of chemical reaction network parameters given reaction degeneracy: an approximate analytic solution"
Although chemical reaction networks (CRN) provide performant and biophysically plausible models for explaining single-cell genomic data, inference of reaction network parameters in this setting usually assumes available data points can be viewed as independent samples from a steady state distribution. Less is known about how to perform efficient parameter inference in the case that there is a continuous-time data stream, which adds complexity like nontrivial correlations between samples from different times. In the continuous-time setting, one has two natural questions: (i) given a set of reactions that could plausibly explain the observed data stream, what are reasonable estimates of the associated reaction rate parameters? and (ii) what is the minimal set of reactions necessary to explain the data? Both questions can be formalized as Bayesian inference problems, with the former concerning the inference of a model-dependent parameter posterior, and the latter concerning ‘structure’ inference. If one can assume each possible reaction has a different stoichiometry vector, there is a well-known analytic solution to both problems; if reactions can have the same stoichiometry vector (i.e., there is reaction degeneracy), both problems become substantially more difficult, and no analytic solution is known. We present the first approximate analytic solution to both problems, which is valid when the number of observations becomes sufficiently large. In its regime of validity, this solution allows its user to avoid expensive likelihood computations that can involve summing over an exponentially large number of terms. We discuss interesting consequences of this solution, like the fact that ‘simpler’ models with fewer reactions are preferred over more complex ones, and the fact that the parameter posteriors of non-identifiable models are strongly prior-dependent.
4. Adelle Coster School of Mathematics & Statistics, UNSW, Sydney Australia
   "Cellular protein transport: Queuing models and parameter estimation in stochastic systems"
Real-world systems, especially in biology, exhibit significant complexity and inherent limitations in observability. What methods can enhance our understanding of the mechanisms underlying their functionality? Additionally, how can we develop and test explanatory models within a stochastic environment? Evaluating the effectiveness of these models requires quantitative measurements of the disparity between model outputs and observed data. While mean-field, deterministic models have well-established approaches for such assessments, stochastic systems—particularly those constrained by multiple data types—need carefully designed quantitative comparison methods. Methods for inferring the parameters of stochastic models generally require analytical forms of the model solutions, large data sets, summary statistics, or assumptions on the distribution of model outputs. These approaches can be limiting if you wish to preserve the information in the variability of the data but you do not have sufficient data to reliably fit distributions or determine robust statistics. We present a hierarchical approach to develop a distance measure for the direct comparison of model output distributions to experimentally observed distributions, avoiding any assumptions about distributions and the need to choose summary statistics. Our distance measure allows for constraining the model with multiple experiments, not necessarily of the same type, such that each experiment constrains some, or all, of the model parameters. We use this distance for parameter estimation with our queuing model of intracellular GLUT4 translocation. We will explore some practical considerations when using the distance for parameter inference, such as the effects of model output sampling and experimental error. Fitting the queuing model to data allowed us to uncover a possible mechanism of GLUT4 sequestration and release in response to insulin. Authors: Brock D. Sherlock and Adelle C.F. Coster
5. Clark Kendrick Go Collaborative Analytics Group, Department of Mathematics, Ateneo de Manila University
   "Exploring Mathematical Techniques in Collective Behaviour and Decision Making in Animal Groups"
Collective behaviour in animal groups are coordinated movements and interactions among members that aim to achieve a common goal. Whether these goals are for allocation of resources or defence from predators, the collective behaviour appears to be largely a group activity initiated by a member, known as the leader. In the absence of high-resolution spatio-temporal data, various qualitative studies offer a glimpse of how leader-follower interactions take place. For example, Nagy, et al., studied the average delay in response when pigeons change the direction inflight. Next, Bourjade, et al., studied the first mover and the succeeding order of movements of Przewalski's Horses. Furthermore, various studies on the collective motion in the animal kingdom offer mathematical models and infer how the interactions and decision making take place. Important questions arise during an event of coordinated motion in animals. During such an event, do individuals move according to a certain set of natural rules? Or certain patterns form due to the influence of a leader? How is this influence measured? Finally, how is influence transferred to other members of the group? In this study, we discuss the role of information theory to quantitatively uncover leader-follower relationship in a horse group. Specifically, we introduce concepts from information theory, specifically global and local transfer entropy being applied to a harem of horses. We will discuss their definitions, and how these key concepts are used to support causation in events. We will then discuss some important implications on how this technique can be used to analyse collective motion where data is scarce.