Chris Teska, Kurt Pasque, Ruriko Yoshida, Baran Hashemi
In this work, we introduce a Tropical Axial Attention neural reasoning architecture that replaces vanilla softmax dot-product attention with max-plus operators, inducing a piecewise-linear structure aligned with dynamic programming formulations. From multi-species sequence alignments, our model learns all possible pairwise distances and is trained using a combination of ℓ 1 and tropical symmetric distance metric losses with an ultrametric violation penalty. We leverage the well known isomorphic relationship between the space of all phylogenetic trees with n species and tropical Grassmannian to show that tropical attention provides a natural geometric framework for phylogenetic inference. On empirical D S 1 − D S 11 alignments, where true trees are unknown, the tropical model achieves the lowest MAE to its FastME-induced tree metric on every dataset, with a MAE reductions averaging 81.5% relative to Phyloformer and 98.4% relative than Phyloformer 2. These results suggest that tropical attention is a useful geometric inductive bias for neural phylogenetic inference, especially under distribution shift and when tree-metric consistency is important.
Elizabeth A. Burke, Samantha N. Schumm, Amy Jobe, Kara Whelply, Ishita Das
The current landscape of science communication is undergoing rapid transformation, shaped by profound societal shifts including the COVID-19 pandemic, the rise of social media, the decline of traditional news media, and the growing influence of large language models. These changes have fundamentally altered how the biomedical community and the public share, consume, and trust scientific information. This article contextualizes these changes, explains why redesigning the science communication ecosystem is urgent, and offers recommendations to improve accuracy, strengthen trust, and deepen public engagement. Specifically, philanthropy can play a key role by driving momentum and optimizing the flow of scientific information across three stages: (1) scientific discovery to publication, (2) interpretation and distribution, and (3) public engagement.
Markov modulated (MM) models in the theory of matrix-analytic methods (MAMs) are a class of probabilistic models whose key feature is an underlying Markov environment that drives the evolution of some complex and unpredictable system. Our goal in this paper is to emphasise their great potential for understanding evolution and the need for close multidisciplinary collaborations between probabilists and evolutionary biologists to advance the field. We give an overview of the fundamental MM models in MAMs, namely quasi-birth-and-death processes (QBDs) and stochastic fluid models (SFMs), in which the fluctuations in the level variable are modulated by fluctuations in the phase variable. We illustrate their application potential in evolution through an example of phylogenetic inference, state a general tree likelihood formula applicable to a wide range of MM models, and discuss future directions required to advance the theory of MM models as well as in the applications of the existing theory to evolutionary contexts.
Federico Surra, Francesco Gonella, Luigi Conte, Alessandra Romano
Biological complex systems self-organize and evolve through feedback-driven interactions not adequately captured by linear cause-effect models. To address this gap, we propose a novel approach grounded in Energy Systems Language (ESL). It provides a description of how resource availability, carrying capacity and feedback networks shape the system trajectories. ESL is rooted in open systems Thermodynamics, shifting the focus from the biology of events to their systemic structure. Models are built on biomedical knowledge, with parameters determined by empirical observation and biophysical constraints, producing simulators that address “what if” exploratory questions. We applied this approach to the antigen-specific adaptive immune response as a proof-of-concept of the framework designed to investigate system-level dynamic regime stability and qualitative response to external driving forces. We developed a computational simulator reproducing hallmark features of humoral immunity at daily, monthly and decadal time-scales, including germinal-center dynamics, establishment of long-lived plasma cells, antibody secretion, and long-term decline in the absence of antigen re-exposure. Using this approach, future investigations of complex pathologies, e.g., lymphoid malignancies and autoimmune diseases, can be developed, providing computational tools for testing hypotheses on patient-specific therapeutic strategies.
The coherence of living systems extends beyond the lifespan of the individual organism: it persists across biological lineages, through the turnover of organisms and change across deep evolutionary time. This paper terms this phenomenon evolutionary coherence, the lineage-scale expression of a more general property, biological coherence, whose organism-scale expression systems biology has characterised while its evolutionary counterpart has not. Biological coherence is analysed through four aspects — stability, transformation, connectivity, and coordination — each expressed at both the lineage and the organism scale. Their lineage-level manifestations appear in familiar but unreconciled patterns: persistent morphologies, punctuated transformation, structured hybridisation, and the channelled forms evolutionary change repeatedly takes. At the organism scale the same aspects govern the maintenance of organised function and its breakdown, including the convergence of distinct systemic diseases on shared signatures of regulatory dysfunction. Such patterns are not readily accommodated by the foundational ideas of evolutionary biology and biomedicine, which engage auxiliary hypotheses to account for them; some point toward higher-order organisation but stop short of articulating it as primary. Proceeding by abductive inference, the paper specifies the requirements a unifying account must meet, then surveys the molecular evidence and develops a systems model of their operation. The architectures examined are hierarchical gene regulatory networks, bow-tie physiological regulation, germline regulatory systems, the regulation of repetitive DNA, and centromere identity. The framework is articulated as the CHORA Hypothesis: that a Common Higher-Order Regulatory Architecture, common to the organism and the lineage, operates as an integrated level across organisms and generations; that the germline is its principal conduit; that the features sustaining coherence also underlie its failure; and that its disruption at the organism level underlies these convergent signatures of dysfunction. The architecture is not a further agency but the largely conserved higher-order organisation realised by established molecular processes — transposable-element activity foremost among them — and through which coherence is sustained. These processes are its molecular signatures; the architecture’s own nature is left open. Health and entrenched disease are, on this account, alternative regimes of the one architecture. Part II develops the discriminating predictions and systemic-health implications.
