TBA

Humans and animals do not always act rationally; they exploit rewards but also explore driven by curiosity. However, the mechanisms behind such curiosity-driven behaviors remain unclear. Here, we developed a decision-making model for a two-choice task based on the free energy principle (FEP), integrating perception and action selection. The model explains irrational behaviors depending on curiosity levels. We also proposed a machine learning method called ‘inverse FEP’ to decode temporal changes in curiosity from behavioral data. Applying it to rat behavior, we found that rats exhibited negative curiosity, preferring certain options, and that curiosity increased with expected future information gain. Our approach provides a tool for studying reward–curiosity conflicts and may aid in diagnosing mental disorders.

Statistical physics approaches to understanding complex systems have been remarkably successful. Numerous scientific fields are undergoing a data revolution, and high accuracy simple models for complex phenomena are in high demand. In particular, the maximum entropy principle as formulated by Jaynes provides a powerful framework to capture the collective nature of an ensemble while keeping all but a few measurement constraints maximally random. Despite their ubiquity, deciding whether these models are ‘successful’ remains a challenge. Here, we study the activity of 1000+ simultaneously active cells in the brain of mice as they navigate a virtual environment. Leveraging the massive scale of these data, we compared 900 different models for different subgroups of 100 neurons each. We demonstrate how employing simple models—analogous to Ising models with competing interactions from physics—enables accurate predictions of population properties. These models are particularly effective when applied to local subgroups of neurons. In contrast, their predictive performance significantly diminishes when constructed for distant neuronal subgroups. Through a systematic comparison across an extensive number of model predictions, we establish a precise scale and language for what constitutes a 'successful’ model.

Chemotactic navigation of biological cells represents a prime example of cellular information processing corrupted by noise, with known input and output, and a single objective. Intriguingly, cells of different size use different chemotaxis strategies, comparing concentrations of signalling molecules in either time or space: Small and fast bacteria sense concentration gradients in time while moving actively, whereas large and slow eukaryotic cells with crawling motility sense concentration gradients across their diameter in space. Only heuristic arguments exist to explain this evolutionary choice. 

To address this open question, I will present an information theory of an ideal agent that combines both gradient-sensing strategies. By quantifying 'chemotaxis in bits', we can compare the information gain of temporal and spatial gradient-sensing. This theory of Bayesian chemotaxis generalizes information-greedy infotaxis to spatially extended agents with motility noise and an egocentric map. 

Using information decomposition, we can predict when each gradient-sensing strategy provides more information as function of a power law that combines agent size, motility noise and sensing noise into a single predictor. This predictor is consistent with data from 250 chemotactic cells, including unusual bacteria that use spatial gradient-sensing. 

Our idealized model of Bayesian chemotaxis assuming unlimited information processing capabilities thus serves as a benchmark for the chemotaxis of biological cells.

Auconi et al., EPL 138, 12001 (2023); Rode et al., PRX Life 2, 023012 (2024)

Autocatalytic sets of molecules are candidates for the most primitive chemical organizations that appeared on the prebiotic Earth. Their emergence may have been a crucial step in the origin of life. An open question about autocatalytic sets pertains to their evolvability and ability to grow more complex in a prebiotic environment. This talk will present a mathematical model that shows how autocatalytic sets can evolve when they are placed within protocells. The molecules that form the boundary of the protocell are assumed to be part of the autocatalytic set, and this boundary is assumed to be permeable to small molecules in the environment. A combination of nonlinear chemical dynamics and stochasticity in small enclosures gives rise to a Darwinian like evolution of this simple physico-chemical system.

"Maintaining stability is a critical issue for living systems. Robust perfect adaptation (RPA) is a control-theoretical mechanism that enables certain output variables to attain and sustain desired values despite external disturbances in a robust manner. RPA helps the survival of living systems in unpredictable environments, and as such there are numerous examples of biological implementations of this feature. However, identifying RPA properties and associated regulatory mechanisms is highly nontrivial problem given the complexity of biological systems. 

In this talk, we elucidate the essential role of network topology in the phenomenon of RPA [1]. We have shown that all the RPA properties in a deterministic chemical reaction system can be identified by topological characteristics of subnetworks. Furthermore, we identify integral feedback controller realize each RPA property, casting our results into the control-theoretic paradigm of the Internal Model Principle. 

Reference:

[1] Yuji Hirono, Ankit Gupta, Mustafa Khammash, [arXiv:2307.07444, PRX Life, in press]"

To maintain information in a molecular system, it must be somehow replicated, but an accurate replication is only possible below a certain length for the molecules that carry the information. Above that threshold, information loss occurs, as shorter molecules called parasites take over the system, preventing information rich molecules to be replicated. Transient compartmentalization is a way to get around this issue, allowing parasites to be controlled and information to be passed along generations. Experiments using compartmentalized RNA molecular systems have confirmed this scenario. When carried on a long time, these experiments have also shown that coevolution of parasites and replicases take place within an ecosystem of increasing complexity. Here we present a theoretical model of transient compartmentalization that accounts for mutations, evolution with a gradual improvement of fitness and co-evolution. 

To double the cellular population of ribosomes, a fraction of the active ribosomes is allocated to synthesize ribosomal proteins. Subsequently, these ribosomal proteins enter the ribosome self-assembly process, synthesizing new ribosomes and forming the well-known ribosome autocatalytic subcycle. Neglecting ribosome lifetime and the duration of the self-assembly process, the doubling rate of all cellular biomass can be equated with the fraction of ribosomes allocated to synthesize an essential ribosomal protein times its synthesis rate. However, ribosomes have a finite lifetime, and the assembly process has a finite duration. Furthermore, the number of ribosomes is known to decrease with slow growth rates. The finite lifetime of ribosomes and the decline in their numbers present a challenge in sustaining slow growth solely through controlling the allocation of ribosomes to synthesize more ribosomal proteins. When the number of ribosomes allocated per mRNA of an essential ribosomal protein is approximately one, the resulting fluctuations in the production rate of new ribosomes increase, causing a potential risk that the actual production rate will fall below the ribosome death rate. Thus, in this regime, a significant risk of extinction of the ribosome population emerges. To mitigate this risk, we suggest that the ribosome translation speed is used as an alternative control parameter, which facilitates the maintenance of slow growth rates with a larger ribosome pool. We clarify the observed reduction in translation speed at harsh environments in E. coli and C. Glutamicum, explore other mitigation strategies and suggest additional falsifiable predictions of our model.

TBA

Collective behavior of animals is a fascinating example of self-organization in biology. This phenomenon is believed to provide several advantages to individuals, such as facilitating exchange of social information, promoting accurate collective decisions, or affording protection from predators. It has been proposed that animal collectives operate near a critical point [1], where collective computations are optimized [1,2]. Recently, we have investigated the "criticality hypothesis" in the context of fish schools responding to predators  [3,4,5,6]. In this talk, I will present our work on so-called "startle cascades", rapid escape responses propagating through fish schools, analogous to neuronal activity avalanches [4,5,6].

[1] T. Mora, W. Biale, J Stat Phys 144, 268-302 (2011);

[2] P. Romanczuk, B.C. Daniels. "Phase transitions and criticality in the collective behavior of animals-self-organization and biological function." In Order, Disorder and Criticality: Advanced Problems of Phase Transition Theory, pp. 179-208. 2023.

[3] P.P. Klamser, P. Romanczuk, PLoS Comp Biol 17, e1008832 (2021);

[4] Sosna et al., PNAS 116.41 (2019);

[5] W. Poel et al., Sci Adv 8, eabm6385 (2022);

[6] L. Gómez Nava et al., Nature Phys 19 (2023);

*This work was supported by the Deutsche Forschungsgemeinschaft (German Research Foundation): RO 4766/2-1 and under Germany's Excellence Strategy - EXC 2002/1 'Science of Intelligence' project- no. 390523135.

Perfect adaptation is a well-studied biochemical homeostatic behavior lying at the core of biochemical regulation. While the concepts of homeostasis and perfect adaptation are not new, their underlying mechanisms and associated biochemical regulation motifs are not yet fully understood. Insights from control theory unraveled the connections between perfect adaptation and integral control, a prevalent engineering control strategy. In particular, the recently introduced Antithetic Integral Controller (AIC) has been shown to successfully ensure perfect adaptation properties to the network it is connected to. The complementary structure of the two molecules the AIC relies upon allows for a versatile way to control biochemical networks, a property which gave rise to an important body of literature pertaining to mathematically elucidating its properties, generalizing its structure, and developing experimental methods for its implementation. The Antithetic Integral Rein Controller (AIRC), an extension of the AIC in which both controller molecules are used for control, holds many promises as it supposedly overcomes certain limitations of the AIC. We focus here on an AIRC with output inhibition that combines two AICs in a single network. We theoretically demonstrate its superior properties, which are linked to the intrinsic properties of the niAIC with output inhibition. This controller ensure structural stability and structural perfect adaptation properties for the controlled network under mild assumptions, meaning that this property is independent of the parameters of the network and the controller. The results are very general and valid for the class of unimolecular mass-action networks as well as more general networks, including cooperative and Michaelis-Menten networks. We also provide a systematic and accessible computational way for verifying whether a given network satisfies the conditions under which the structural property would hold.

Although the present translation system has high fidelity and 20 standard amino acids alphabet, early life had a genetic code with low fidelity and a smaller number of amino acids. What would have been the outcome of protein evolution at the early stages of life?  For directed evolution with such codes, we have constructed artificial genetic codes with a reduced alphabet and codes with low fidelity.  Adjustable misincorporation by our low-fidelity codes can reduce the risk of trapping at local optima during artificial evolution by smoothing the protein fitness landscape in the DNA sequence space. Furthermore, modification of the fundamental system of life provides novel engineering tools and new insights into life.

We have developed a Probabilistic Machine-Learning (ML) Framework, based on Physics-Informed Gaussian Process Regression, that is able to infer solutions to arbitrary Stokes Flow problems given sparse and noisy data. We refer to the method as Stokesian Process (SP). In contrast to more popular ML approaches, e.g., based on Physics-Informed Neural Networks or Neural Operators, our SP method is able to exactly encode the governing equations (i.e., Stokes and continuity equations) into the inference, resulting in more efficient training and more robust predictions. The SP method will be of particular use to analyze experiments, e.g., Particle Image Velocimetry data for biological flows.

A key feature of many developmental systems is their ability to self-organize spatial patterns of functionally distinct cell fates. A spectacular example of this ability are artificial stem cell assemblies, which are paving the way towards a quantifiable self-organization of biological systems. To ensure proper biological function, such patterns must be established reproducibly, by controlling and even harnessing intrinsic and extrinsic fluctuations. However, conceptual theoretical frameworks to understand how cell-cell communication and tissue heterogeneity result in multicellular self-organization are lacking. To address this gap, we develop a dynamical systems model for intestinal organoid self-organization. This event is driven by Dll1/Notch lateral inhibition, but also requires heterogeneous activity of Yap1. We develop a biophysical model of collective cellular decisions, combining lateral inhibition, heterogeneous cellular states, and active epithelial mechanics. Using a set of perturbation experiments including Yap1 inhibition and overexpression to constrain the model, we make key quantitative predictions for cellular states during symmetry-breaking, including a prepatterned Dll1 state based on Yap1 activity. By combining high-throughput imaging and single-cell omics, we confirm our predictions and reveal how cells integrate the cell- and tissue-scale signals into the prepatterned Dll1 state via FoxA transcription factors, through extensive epigenetic remodeling of secretory progenitor cells. Taken together, we demonstrate how minimal biophysical models can make key predictions for stem cell fate choices in complex multicellular systems. This approach reveals how cells sense and integrate tissue-level information to take the first cell fate decision, break symmetry and properly pattern intestinal organoids.

"The emergence and persistence of polymorphism within populations generally requires specific selective regimes. Here, we develop an unifying theoretical framework to explore how disassortative mating can generate and maintain polymorphism at the targeted loci.  Our theoretical study  confirms that the conditions for the persistence of a given level of allelic polymorphism depend on the relative reproductive advantages among pairs of individuals. Interestingly, equilibria with unbalanced allelic frequencies were shown to emerge from successive introduction of mutants.We then investigate the role of the function linking allelic divergence to reproductive advantage on the evolutionary fate of alleles within population. Our results highlight the significance of the shape of this function on both the number of alleles maintained and their level of genetic divergence.

Joint work with C. Coron, H. Leman, V. Llaurens and C. Smadi (https://link.springer.com/article/10.1007/s00285-022-01832-1)"

"Human behaviour such as social distancing or quarantining of infected individuals can affect the spread of diseases. In models, individuals are typically treated either as passive agents which do not modify their behaviour in response to the disease or their behavioural response is assumed to be motivated by self-interest. In such frameworks, people show little concern for others and a significant reduction in social activity, such as quarantining of infected people, requires government enforcement. However, people are likely to be at least weakly altruistic and thus interested in protecting others. Unfortunately, little is known theoretically about how altruism affects the course of epidemics. 

Here, we show that even extremely weakly altruistic populations can be expected to social distance when infected with a dangerous disease in a self-organised way, thus substantially suppressing epidemic spread. 

In a game theoretic extension of a simple compartmental disease model, we assume that well-informed, rational individuals seek to minimise both their own costs, coming from infection and social distancing, as well as the corresponding costs borne by the population. The strength of individual altruism is given by χ, where an individual weights the welfare of 1/χ other population members as equal to their own. We find that infected individuals can spontaneously suppress the disease through social distancing, even at remarkably low levels of altruism. This outcome remains rational even in the presence of a moderate level of asymptomatic cases and/or completely selfish individuals."

Understanding biological functions requires identifying the physical limits and system-specific constraints that have shaped them. In Escherichia coli chemotaxis, gradient-climbing speed is information-limited, bounded by the sensory information they acquire from real-time measurements of their environment. However, it remains unclear what limits this information. Past work conjectured that E. coli’s chemosensing is limited by the physics of molecule arrivals at their sensors. Here, we derive the physical limit on behaviorally-relevant information, and then perform single-cell experiments to quantify how much information E. coli’s signaling pathway encodes. We find that E. coli encode two orders of magnitude less information than the physical limit due to their stochastic signal processing. Thus, system-specific constraints, rather than the physical limit, have shaped the evolution of this canonical sensory-motor behavior.

Biophysical constraints limit the specificity with which transcription factors (TFs) can target regulatory DNA. While individual nontarget binding events may be low affinity, the sheer number of such interactions could present a challenge for gene regulation by degrading its precision or possibly leading to an erroneous induction state. We first review our past work on how detrimental such nontarget binding (“crosstalk”) is in the context of equilibrium gene regulatory schemes that have proven to be good models of regulation for prokaryotes. We briefly review some suggestions for how non-equilibrium regulatory schemes could implement kinetic proofreading to circumvent the equilibrium constraints. The focus of this presentation is our recent work on how a particular non-equilibrium mechanism that involves chromatin remodeling can endow gene regulatory networks with precise regulation, despite nontarget binding.

Complex systems theory teaches us that simple, higher-level laws with few effective parameters often emerge from the interaction of small-scale components. As biology is becoming more and more quantitative, one can use a combination of first-principle theoretical modeling with simple machine-learning techniques to build accurate and tractable theories of biological dynamics, best understood in low-dimensional latent spaces. I will focus on the problem of immune recognition by T cells. Using the Immunotron robotic platform developed with the Altan-Bonnet group (US National Cancer Institute), we generate high-dimensional dynamical data (cytokines and dozens of cell surface markers) for T cell's immune response. Using a combination of information theory and machine learning, we compute an interpretable latent representation of T cell dynamics, build decoding maps, and characterize modes of T cells’ activation.  Remarkably, the latent space topography naturally accounts for non-trivial features of the immune response, such as positive selection and immune antagonism. This suggests that geometric information directly related to important biological information can be extracted from high-throughput data, and used for theoretical modeling and applications.

Comprehending cell death is one of the central topics of biological science. Currently, the criteria for microbial cell death are purely experimental, based on PI staining and regrowth experiments. In the present project, we aimed to develop a mathematical framework of cell death based on the metabolic state of the cells. 

Our attempt is to develop a theoretical framework of ""death"" for cellular metabolism. We start by defining dead states as cellular metabolic states that are not returnable to the predefined ""representative living states"" regardless of the modulation of enzyme concentrations and external nutrient concentrations. The definition requires a method to compute the restricted, global, and nonlinear controllability, for which no general theory exists. We have developed ""The Stoichiometric Rays"", a simple method to solve the controllability computation. This allows us to compute how the enzyme concentration should be modulated to control the metabolic state from a given state to a desired state. 

Using the stoichiometric rays, we have computed the returnability of the non-growing state emerging in an in silico metabolic model of E. coli to the growing state of the model, and found that the non-growing state is indeed a ""dead"" state. Furthermore, we have quantified “the Separating Alive and Non-life Zone (SANZ) hypersurface” which divides the phase space into the living- and non-living regions.   

In this talk, I would like to present our framework for cell death, including stoichiometric rays, and what we can learn from quantifying the SANZ hypersurface.  

TBA