More Like this

1. Optimal phenotypic adaptation in fluctuating environments

   https://doi.org/10.1016/j.bpj.2023.10.019  

   George, Jason T (November 2023, Biophysical Journal)

   Phenotypic adaptation is a universal feature of biological systems navigating highly variable environments. Recent empirical data support the role of memory-driven decision making in cellular systems navigating uncertain future nutrient landscapes, wherein a distinct growth phenotype emerges in fluctuating conditions. We develop a simple stochastic mathematical model to describe memory-driven cellular adaptation required for systems to optimally navigate such uncertainty. In this framework, adaptive populations traverse dynamic environments by inferring future variation from a memory of prior states, and memory capacity imposes a fundamental trade-off between the speed and accuracy of adaptation to new fluctuating environments. Our results suggest that the observed growth reductions that occur in fluctuating environments are a direct consequence of optimal decision making and result from bet hedging and occasional phenotypic-environmental mismatch. We anticipate that this modeling framework will be useful for studying the role of memory in phenotypic adaptation, including in the design of temporally varying therapies against adaptive systems. 

   more » « less
2. Online Prediction in Sub-linear Space

   Peng, Binghui; Zhang, Fred (January 2023, Proceedings of the 2023 Annual ACM-SIAM Symposium on Discrete Algorithms (SODA))

   We provide the first sub-linear space and sub-linear regret algorithm for online learning with expert advice (against an oblivious adversary), addressing an open question raised recently by Srinivas, Woodruff, Xu and Zhou (STOC 2022). We also demonstrate a separation between oblivious and (strong) adaptive adversaries by proving a linear memory lower bound of any sub-linear regret algorithm against an adaptive adversary. Our algorithm is based on a novel pool selection procedure that bypasses the traditional wisdom of leader selection for online learning, and a generic reduction that transforms any weakly sub-linear regret o(T) algorithm to T1-α regret algorithm, which may be of independent interest. Our lower bound utilizes the connection of no-regret learning and equilibrium computation in zero-sum games, leading to a proof of a strong lower bound against an adaptive adversary. 

   more » « less
3. Mapping the dynamics of epigenetic adaptation in S. pombe during heterochromatin misregulation

   https://doi.org/10.1016/j.devcel.2024.07.006  

   Larkin, Ajay; Kunze, Colin; Seman, Melissa; Levashkevich, Alexander; Curran, Justin; Morris-Evans, Dionysus; Lemieux, Sophia; Khalil, Ahmad S; Ragunathan, Kaushik (August 2024, Developmental Cell)

   Epigenetic mechanisms enable cells to develop novel adaptive phenotypes without altering their genetic blueprint. Recent studies show histone modifications, such as heterochromatin-defining H3K9 methylation (H3K9me), can be redistributed to establish adaptive phenotypes. We developed a precision-engineered genetic approach to trigger heterochromatin misregulation on-demand in fission yeast. This enabled us to trace genome-scale RNA and H3K9me changes over time in long-term, continuous cultures. Adaptive H3K9me establishes over remarkably slow timescales relative to the initiating stress. We captured dynamic H3K9me redistribution events which depend on an RNA binding complex MTREC, ultimately leading to cells converging on an optimal adaptive solution. Upon stress removal, cells relax to new transcriptional and chromatin states, establishing memory that is tunable and primed for future adaptive epigenetic responses. Collectively, we identify the slow kinetics of epigenetic adaptation that allow cells to discover and heritably encode novel adaptive solutions, with implications for drug resistance and response to infection. 

   more » « less
4. Links between Innate and Adaptive Immunity Can Favor Evolutionary Persistence of Immunopathology

   https://doi.org/10.1093/icb/icae105  

   Cressler, Clayton_E; Adelman, James_S (July 2024, Integrative And Comparative Biology)

   Synopsis Immunopathology, or the harm caused to an organism’s own tissues during the activation of its immune system, carries substantial costs. Moreover, avoiding this self-harm may be an important mechanism underlying tolerance of infection, helping to reducing fitness costs without necessarily clearing parasites. Despite the apparent benefits of minimizing immunopathology, such damage persists across a range of host species. Prior work has explored a trade-off with resistance during a single infection as a potential driver of this persistence, with some collateral damage being unavoidable when killing parasites. Here, we present an additional trade-off that could favor the continued presence of immunopathology: robust immune responses during initial infection (e.g., innate immunity in vertebrates) can induce stronger memory (adaptive immunity), offering protection from future infections. We explore this possibility in an adaptive dynamics framework, using theoretical models parameterized from an ecologically relevant host-parasite system, house finches (Haemorhous mexicanus) infected with the bacterial pathogen, Mycoplasma gallisepticum. We find that some degree of immunopathology is often favored when immunopathology during first infection either reduces susceptibility to or enhances recovery from second infection. Further, interactions among factors like transmission rate, recovery rate, background mortality, and pathogen virulence also shape these evolutionary dynamics. Most notably, the evolutionary stability of investment in immunopathology is highly dependent upon the mechanism by which hosts achieve secondary protection (susceptibility vs. recovery), with the potential for abrupt evolutionary shifts between high and low investment under certain conditions. These results highlight the potential for immune memory to play an important role in the evolutionary persistence of immunopathology and the need for future empirical research to reveal the links between immunopathology during initial infections and longer-term immune protection. 

   more » « less
5. Reason for Future, Act for Now: A Principled Architecture for Autonomous LLM Agents

   Liu, Zhihan; Hu, Hao; Zhang, Shenao; Guo, Hongyi; Ke, Shuqi; Liu, Boyi; Wang, Zhaoran (October 2024, ICML)

   Large language models (LLMs) demonstrate impressive reasoning abilities, but translating reasoning into actions in the real world remains challenging. In particular, it is unclear how to complete a given task provably within a minimum number of interactions with the external environment, e.g., through an internal mechanism of reasoning. To this end, we propose the first framework with provable regret guarantees to orchestrate reasoning and acting, which we call “reason for future, act for now” (RAFA). Specifically, we design a prompt template for reasoning that learns from the memory buffer and plans a future trajectory over a long horizon (“reason for future”). At each step, the LLM agent takes the initial action of the planned trajectory (“act for now”), stores the collected feedback in the memory buffer, and reinvokes the reasoning routine to replan the future trajectory from the new state. The key idea is to cast reasoning in LLMs as learning and planning in Bayesian adaptive Markov decision processes (MDPs). Correspondingly, we prompt LLMs with the memory buffer to estimate the unknown environment (learning) and generate an optimal trajectory for multiple future steps that maximize a value function (planning). The learning and planning subroutines are performed in an “incontext” manner to emulate the actor-critic update for MDPs. Our theoretical analysis establishes a √T regret, while our experimental validation demonstrates superior empirical performance. Here, T denotes the number of online interactions. 

   more » « less