Simon Schnyder
John Jairo Molina Lopez
Gašper Tkačik
David Brückner
Matthew S. Turner
The aim of this session is to discuss how complex biological phenomena from embryonic development to the collective motion of organisms can be understood via optimisation principles.
9:30-10:30: Talk by Gašper Tkačik
10:30-11:00: Break
11:00-12:00: Talk by David Brückner
12:00-13:00: Lunch
13:00-14:00: Talk by Matthew S. Turner
18:00-21:00: Dinner (taxi leave at 18:00 from Lab5 Parking Lot)
Many biological systems operate near the physical limits to their performance, suggesting that aspects of their behavior and underlying mechanisms could be derived from optimization principles. However, such principles have often been applied only in simplified models. Here, we explore a detailed mechanistic model of the gap gene network in the Drosophila embryo, optimizing its 50+ parameters to maximize the information that gene expression levels provide about nuclear positions. This optimization is conducted under realistic constraints, such as limits on the number of available molecules. Remarkably, the optimal networks we derive closely match the architecture and spatial gene expression profiles observed in the real organism. Our framework quantifies the trade-offs involved in maximizing functional performance and allows for the exploration of alternative network configurations, addressing the question of which features are necessary and which are contingent. Our results suggest that multiple solutions to the optimization problem might exist across closely related organisms, offering new insights into the evolution of gene regulatory networks.
Embryonic development is a spectacular display of self-organization of multi-cellular systems, combining transformations of tissue mechanics and patterns of gene expression. These processes are driven by the ability of cells to communicate through mechanical and chemical signaling, allowing coordination of both collective movement and patterning of cellular states. To ensure proper biological function, such patterns must be established reproducibly, by controlling and even harnessing intrinsic and extrinsic fluctuations. While the relevant molecular processes are increasingly well understood, we lack principled frameworks to understand how tissues obtain information to generate reproducible patterns. I will present an information-theoretic framework to mathematically define and interpret the reproducibility and robustness of fate patterns. This framework provides a normative approach for optimization of cell signaling and mechanics, which I showcase using a variety of mechanistic models ranging from reaction-diffusion systems to delta-notch signaling. Furthermore, our approach allows us to estimate the reproducibility of experimental gene expression profiles of stem cell gastruloids, a self-organizing in vitro model of mammalian development. We demonstrate that these estimators assess the total amount of information contained in these expression profiles. This information is, in principle, accessible to the cells by local or non-local readouts, and provides a generalization of the concept of positional information, which is strictly local. Our work opens up an avenue towards unifying the zoo of chemical and mechanical signaling processes encountered across different developmental systems by using a common information-theoretic language.
Recent work has shown that a "bottom up" model of collective motion can be derived starting from the principle that agents seek to maximise the space of environmental states accessible in the future (FSM). This can be shown to generate dynamics that preserve a cohesive collective. This collective can either be highly aligned (like birds) or with low orientational order (like insects) depending on the agents’ speed. This has further allowed us to construct a neural network-based model that generates similar collective motion using purely visual input [Charlesworth & Turner, PNAS 2019]. Such models are more realistic candidates for animal behaviour than "top down" models, like variants of the one proposed in [Vicsek et al., PRL 1995], that are known to have various unphysical features. We will introduce the FSM model and discuss recent extensions that allow us to assign additional preferences to some (or all) agents. In this way we can construct hybrid models that retain the strength of future state maximisation with, e.g. spatial destinations preferred by some informed subset of agents.