In neuroscience, joint receptors have traditionally been viewed as limit detectors, providing positional information only at extreme joint angles, while muscle spindles are considered the primary sensors of joint angle position. However, joint receptors are widely distributed throughout the joint capsule, and their full role in proprioception remains unclear. In this study, we specifically focused on mimicking Type I joint receptors, which respond to slow and sustained movements, and quantified their proprioceptive potential using a biomimetic joint developed with robotics technology. Results showed that Type I-like joint receptors alone enabled proprioceptive sensing with an average error of less than 2 degrees in both bending and twisting motions. These findings suggest that joint receptors may play a greater role in proprioception than previously recognized and that the relative contributions of muscle spindles and joint receptors are differentially weighted within neural networks during development and evolution. Furthermore, this work may prompt new discussions on the differential proprioceptive deficits observed between the elbows and knees in patients with hereditary sensory and autonomic neuropathy type III. Together, these findings highlight the potential of biomimetics-based robotic approaches for advancing interdisciplinary research bridging neuroscience, medicine, and robotics.

CVTree is an alignment-free methodology for inferring species phylogeny and taxonomy. This method allows for the efficient and accurate resolution of evolutionary relationships among large numbers of species based on whole-genome sequence data. Since 2004, we have been continuously providing CVTree web services. Recently, the server has undergone a significant upgrade, culminating in the release of the WebCVTree4 platform. This upgrade encompasses a comprehensive update of the inbuilt genomic database. Concurrently, the core algorithm has been optimized to support online phylogenetic reconstruction for tens of thousands of species, thereby facilitating the construction of genome-based trees of life. Moreover, we have developed a novel algorithm for comparing phylogenetic trees with established taxonomic systems. This algorithm allows for rapid tree rooting, taxonomic annotation, and topology comparison. Through an interactive web-based visualization tool, users can dynamically adjust tree layouts and export high-quality phylogenetic tree figures. This functionality provides robust support for comparative analysis between CVTree-generated phylogeny and taxonomy. As genome sequencing costs continue to decline, research into microbial evolution and the revision of taxonomic frameworks will increasingly rely on whole-genome data. WebCVTree4 will serve as an efficient web-based platform to support studies in microbial phylogenetics and taxonomy, accessible at https://cvtree.online/.

Extrachromosomal DNA (ecDNA) represents one of the most pressing challenges in cancer biology: circular DNA structures that amplify oncogenes, evade targeted therapies, and drive tumor evolution in ~30% of aggressive cancers. Despite its clinical importance, computational ecDNA research has been built on broken foundations. We discover that existing benchmarks suffer from circular reasoning -- models trained on features that already require knowing ecDNA status -- artificially inflating performance from AUROC 0.724 to 0.967. We introduce ECLIPSE, the first methodologically sound framework for ecDNA analysis, comprising three modules that transform how we predict, model, and target these structures. ecDNA-Former achieves AUROC 0.812 using only standard genomic features, demonstrating for the first time that ecDNA status is predictable without specialized sequencing, and that careful feature curation matters more than complex architectures. CircularODE captures ecDNA's unique stochastic dynamics through physics-constrained neural SDEs, achieving r > 0.997 on experimental data via zero-shot transfer. VulnCausal applies causal inference to identify therapeutic vulnerabilities, achieving 80x enrichment over chance and 3.7x higher validation than standard approaches by filtering spurious correlations. Together, these modules establish rigorous baselines for an emerging application area and reveal a broader lesson: in high-stakes biomedical ML, methodological rigor -- eliminating leakage, encoding domain physics, addressing confounding -- outweighs architectural innovation. ECLIPSE provides both the tools and the template for principled computational oncology.

Genomic language models (gLMs) have transformed computational biology, achieving state-of-the-art performance across genomic tasks. Yet a fundamental question threatens the foundation of this success: do these models learn the mechanistic principles governing gene regulation, or do they merely exploit statistical shortcuts? We introduce the Mechanistic Invariance Test (MIT), a rigorous 650-sequence benchmark across 8 classes with scrambled controls that enables clean discrimination between compositional sensitivity and genuine positional understanding. We evaluate five gLMs spanning all major architectural paradigms (autoregressive, masked, and bidirectional state-space models) and uncover a universal failure mode. Through systematic mechanistic probing via AT titration, positional ablation, spacing perturbation, and strand orientation tests, we demonstrate that apparent compensation sensitivity is driven entirely by AT content correlation (r=0.78-0.96 across architectures), not positional regulatory logic. The failures are striking: Evo2-1B and Caduceus score regulatory elements at incorrect positions higher than correct positions, inverting biological reality. All models are strand-blind. Compositional effects dominate positional effects by 46-fold. Perhaps most revealing, a simple 100-parameter position-aware PWM achieves perfect performance (CSS=1.00, SCR=0.98), exposing that billion-parameter gLMs fail not from insufficient capacity but from fundamentally misaligned inductive biases. Larger models show stronger compositional bias, demonstrating that scale amplifies rather than corrects this limitation. These findings reveal that current gLMs capture surface statistics while missing the positional grammar essential for gene regulation, demanding architectural innovation before deployment in synthetic biology, gene therapy, and clinical variant interpretation.

Spiking reservoir computing provides an energy-efficient approach to temporal processing, but reliably tuning reservoirs to operate at the edge-of-chaos is challenging due to experimental uncertainty. This work bridges abstract notions of criticality and practical stability by introducing and exploiting the robustness interval, an operational measure of the hyperparameter range over which a reservoir maintains performance above task-dependent thresholds. Through systematic evaluations of Leaky Integrate-and-Fire (LIF) architectures on both static (MNIST) and temporal (synthetic Ball Trajectories) tasks, we identify consistent monotonic trends in the robustness interval across a broad spectrum of network configurations: the robustness-interval width decreases with presynaptic connection density $β$ (i.e., directly with sparsity) and directly with the firing threshold $θ$. We further identify specific $(β, θ)$ pairs that preserve the analytical mean-field critical point $w_{\text{crit}}$, revealing iso-performance manifolds in the hyperparameter space. Control experiments on Erdős-Rényi graphs show the phenomena persist beyond small-world topologies. Finally, our results show that $w_{\text{crit}}$ consistently falls within empirical high-performance regions, validating $w_{\text{crit}}$ as a robust starting coordinate for parameter search and fine-tuning. To ensure reproducibility, the full Python code is publicly available.

Recent advances in large language models (LLMs) have enabled molecular reasoning for property prediction. However, toxicity arises from complex biological mechanisms beyond chemical structure, necessitating mechanistic reasoning for reliable prediction. Despite its importance, current benchmarks fail to systematically evaluate this capability. LLMs can generate fluent but biologically unfaithful explanations, making it difficult to assess whether predicted toxicities are grounded invalid mechanisms. To bridge this gap, we introduce ToxReason, a benchmark grounded in the Adverse Outcome Pathway (AOP) that evaluates organ-level toxicity reasoning across multiple organs. ToxReason integrates experimental drug-target interaction evidence with toxicity labels, requiring models to infer both toxic outcomes and their underlying mechanisms from Molecular Initiating Event (MIE) to Adverse Outcome (AO). Using ToxReason, we evaluate toxicity prediction performance and reasoning quality across diverse LLMs. We find that strong predictive performance does not necessarily imply reliable reasoning. Furthermore, we show that reasoning-aware training improves mechanistic reasoning and, consequently, toxicity prediction performance. Together, these results underscore the necessity of integrating reasoning into both evaluation and training for trustworthy toxicity modeling.

Contextual clinical reasoning demands robust inference grounded in complex, heterogeneous clinical records. While state-of-the-art fine-tuning, in-context learning (ICL), and retrieval-augmented generation (RAG) enable knowledge exposure, they often fall short of genuine contextual internalization: dynamically adjusting a model's internal representations to the subtle nuances of individual cases at inference time. To address this, we propose Dual-Stream Calibration (DSC), a test-time training framework that transcends superficial knowledge exposure to achieve deep internalization during inference. DSC facilitates input internalization by synergistically aligning two calibration streams. Unlike passive context exposure, the Semantic Calibration Stream enforces a deliberative reflection on core evidence, internalizing semantic anchors by minimizing entropy to stabilize generative trajectories. Simultaneously, the Structural Calibration Stream assimilates latent inferential dependencies through an iterative meta-learning objective. By training on specialized support sets at test-time, this stream enables the model to bridge the gap between external evidence and internal logic, synthesizing fragmented data into a coherent response. Our approach shifts the reasoning paradigm from passive attention-based matching to an active refinement of the latent inferential space. Validated against thirteen clinical datasets, DSC demonstrates superiority across three distinct task paradigms, consistently outstripping state-of-the-art baselines ranging from training-dependent models to test-time learning frameworks.

Why do we forget? Why do we remember things that never happened? The conventional answer points to biological hardware. We propose a different one: geometry. Here we show that high-dimensional embedding spaces, subjected to noise, interference, and temporal degradation, reproduce quantitative signatures of human memory with no phenomenon-specific engineering. Power-law forgetting ($b = 0.460 \pm 0.183$, human $b \approx 0.5$) arises from interference among competing memories, not from decay. The identical decay function without competitors yields $b \approx 0.009$, fifty times smaller. Time alone does not produce forgetting in this system. Competition does. Production embedding models (nominally 384--1{,}024 dimensions) concentrate their variance in only ${\sim}16$ effective dimensions, placing them deep in the interference-vulnerable regime. False memories require no engineering at all: cosine similarity on unmodified pre-trained embeddings reproduces the Deese--Roediger--McDermott false alarm rate ($0.583$ versus human ${\sim}0.55$) with zero parameter tuning and no boundary conditions. We did not build a false memory system. We found one already present in the raw geometry of semantic space. These results suggest that core memory phenomena are not bugs of biological implementation but features of any system that organizes information by meaning and retrieves it by proximity.

Studying learning-related plasticity is central to understanding the acquisition of complex skills, for example learning to master a musical instrument. Over the past three decades, conventional group-based functional magnetic resonance imaging (fMRI) studies have advanced our understanding of how humans' neural representations change during skill acquisition. However, group-based fMRI studies average across heterogeneous learners and often rely on coarse pre- versus post-training comparisons, limiting the spatial and temporal precision with which neural changes can be estimated. Here, we outline an individual-specific precision approach that tracks neural changes within individuals by collecting high-quality neuroimaging data frequently over the course of training, mapping brain function in each person's own anatomical space, and gathering detailed behavioral measures of learning, allowing neural trajectories to be directly linked to individual learning progress. Complementing fMRI with mobile neuroimaging methods, such as functional near-infrared spectroscopy (fNIRS), will enable researchers to track plasticity during naturalistic practice and across extended time scales. This multi-modal approach will enhance sensitivity to individual learning trajectories and will offer more nuanced insights into how neural representations change with training. We also discuss how findings can be generalized beyond individuals, including through statistical methods based on replication in additional individuals. Together, this approach allows researchers to design highly informative longitudinal training studies that advance a personalized account of skill learning in the human brain.

Background: The locus of the gene PTK2B encoding the tyrosine kinase Pyk2 has been associated with the risk of late-onset Alzheimer's disease, the predominant form of dementia. Pyk2 is primarily expressed in neurons where it is involved in excitatory neurotransmission and synaptic functions. Although previous studies have implicated Pyk2 in amyloid-beta and Tau pathologies of Alzheimer's disease, its exact role remains unresolved, with evidence showing both detrimental and protective effects in mouse models. Here, we investigate the role of Pyk2 in hippocampal hyperactivity, Tau synaptic localization and synaptic loss associated with Alzheimer's disease-related alterations occurring in the early stages of the disease. Methods: Pyk2's involvement in amyloid-beta oligomer-induced hippocampal neuronal hyperactivity was investigated using whole-cell patch clamp in hippocampal slices from WT and Pyk2 KO mice. Various Pyk2 mutants were overexpressed in cultured cortical neurons to study Pyk2's role in synaptic loss. Pyk2 and Tau interaction was assessed with bimolecular fluorescence complementation assays in cultured neurons and co-immunoprecipitation in mouse cortex. To evaluate the impact of Pyk2 on Tau expression in synapses, cellular fractionation was performed on hippocampi from WT and Pyk2 KO mice. Results: Genetic deletion of Pyk2 prevented amyloid-beta oligomer-induced hippocampal neuronal hyperactivity and synaptic loss. Overexpression of Pyk2 in neurons decreased dendritic spine density independently of its autophosphorylation or kinase activity, but through its proline-rich motif 1. Furthermore, Pyk2 interacted with Tau in synapses, while Pyk2 deletion decreased Tau synaptic localization in the hippocampus. Conclusions: Pyk2 contributes to hippocampal neuronal hyperactivity and synaptic loss, two early events in Alzheimer's disease pathogenesis. It is also involved in Tau synaptic localization, a process known to be detrimental in Alzheimer's disease. These findings highlight Pyk2 as a critical player in Alzheimer's disease pathophysiology and suggest its potential as a promising therapeutic target for early intervention.

We propose MAE-SAM2, a novel foundation model for retinal vascular leakage segmentation on fluorescein angiography images. Due to the small size and dense distribution of the leakage areas, along with the limited availability of labeled clinical data, this presents a significant challenge for segmentation tasks. Our approach integrates a Self-Supervised learning (SSL) strategy, Masked Autoencoder (MAE), with SAM2. In our implementation, we explore different loss functions and conclude a task-specific combined loss. Extensive experiments and ablation studies demonstrate that MAE-SAM2 outperforms several state-of-the-art models, achieving the highest Dice score and Intersection-over-Union (IoU). Compared to the original SAM2, our model achieves a $5\%$ performance improvement, highlighting the promise of foundation models with self-supervised pretraining in clinical imaging tasks.

Respiration is an essential involuntary function necessary for survival. This poses a challenge for the control of breathing. The preBötzinger complex (preBötC) is a heterogeneous neuronal network responsible for driving the inspiratory rhythm. While neuromodulators such as norepinephrine (NE) allow it to be both robust and flexible for all living beings to interact with their environment, the basis for how neuromodulation impacts neuron-specific properties remains poorly understood. In this work, we examine how NE influences different preBötC neuronal subtypes by modeling its effects through modulating two key parameters: calcium-activated nonspecific cationic current gating conductance ($g_{\rm CAN}$) and inositol-triphosphate ($\rm IP_3$), guided by experimental studies. Our computational model captures the experimentally observed differential effects of NE on distinct preBötC bursting patterns. We show that this dual mechanism is critical for inducing conditional bursting and identify specific parameter regimes where silent neurons remain inactive in the presence of NE. Furthermore, using methods of dynamical systems theory, we uncover the mechanisms by which NE differentially modulates burst frequency and duration in NaP-dependent and CAN-dependent bursting neurons. These results align well with previously reported experimental findings and provide a deeper understanding of cell-specific neuromodulatory responses within the respiratory network.

In both machine learning and in computational neuroscience, plasticity in functional neural networks is frequently expressed as gradient descent on a cost. Often, this imposes symmetry constraints that are difficult to reconcile with local computation, as is required for biological networks or neuromorphic hardware. For example, wake-sleep learning in networks characterized by Boltzmann distributions assumes symmetric connectivity. Similarly, the error backpropagation algorithm is notoriously plagued by the weight transport problem between the representation and the error stream. Existing solutions such as feedback alignment circumvent the problem by deferring to the robustness of these algorithms to weight asymmetry. However, they scale poorly with network size and depth. We introduce spike-based alignment learning (SAL), a complementary learning rule for spiking neural networks, which uses spike timing statistics to extract and correct the asymmetry between effective reciprocal connections. Apart from being spike-based and fully local, our proposed mechanism takes advantage of noise. Based on an interplay between Hebbian and anti-Hebbian plasticity, synapses can thereby recover the true local gradient. This also alleviates discrepancies that arise from neuron and synapse variability -- an omnipresent property of physical neuronal networks. We demonstrate the efficacy of our mechanism using different spiking network models. First, SAL can significantly improve convergence to the target distribution in probabilistic spiking networks versus Hebbian plasticity alone. Second, in neuronal hierarchies based on cortical microcircuits, SAL effectively aligns feedback weights to the forward pathway, thus allowing the backpropagation of correct feedback errors. Third, our approach enables competitive performance in deep networks using only local plasticity for weight transport.

Consciousness science faces the challenge of bridging first-person experience with third-person empirical measurements. Neurophenomenology aims to build such `generative passages' connecting the content of experience with behavioural and neuroscientific data. However, the mathematical machinery for such bridges remains underdeveloped. Here we develop a Rosetta Stone hypothesis from predictive processing, where beliefs serve as a central hub connecting phenomenology, behaviour, and neural dynamics. This hinges on a central technical assumption that phenomenology is a function of beliefs. We pursue a conditional approach: if this assumption holds, then certain predictions mathematically follow. We derive predictions for subjective similarity judgements, cognitive metabolic cost, subjective cognitive effort, and time perception. We review the connection between beliefs and neural dynamics to complete the generative passage for neurophenomenology, omitting the connection between beliefs and behaviour as this is already well-documented elsewhere. Testing our predictions will inform the validity of the central assumption connecting beliefs and phenomenology, and advance the neurophenomenology research programme.

Mutualistic networks provide a powerful way to describe and analyse plant-pollinator communities and their structure over time. While these networks capture the complex interdependencies that link population fates across the season, they can be hard to untangle, preventing us from understanding the emergence of community-scale properties and responses to perturbation. Here, we address this problem by developing a structural model of a plant-pollinator community that explicitly incorporates seasonal turnover and the temporal nature of species interactions. We analyse our model using percolation methods from network science to derive simple analytical solutions linking network structure to emergent community diversity. Our findings reveal that temporal structure organises community diversity into distinct ecological phases, creating the potential for alternative high- and low-diversity states and bistable regimes. We demonstrate how this temporal structure mediates the nature of transitions between these states, determining whether systems undergo gradual shifts or abrupt, catastrophic collapses. Crucially, we show how this temporal structure reduces the robustness of plant-pollinator systems, creating bottlenecks that inhibit species persistence and increase susceptibility to secondary extinctions. Our results demonstrate that the temporal dynamics of plant-pollinator networks are central to mediating their fragility, highlighting the importance of accounting for time when considering community resilience.

Evolutionary game theory has provided substantial contributions to explain the emergence of cooperation under unfavourable conditions in ecology, economics, and the social sciences. Recently, inspired by newly available empirical evidence on group interactions, higher-order networks have emerged as a natural framework to properly encode multiplayer games in structured populations. Here, we study the emergence of cooperation in a nonlinear public goods game (PGG) on hypergraphs, where collective reinforcement captures the synergistic or discounting effect associated with each additional cooperator. In well-mixed populations, single-order PGGs, where all games have the same number of players, display a change in the nature of transition from continuous to discontinuous depending on the exact form of nonlinearity. By contrast, mixed-order PGGs, where games with different number of players coexist, exhibit a richer dynamical regime wherein a state of active coexistence of bistability and cooperation can arise. We further find that scale-free hypergraphs promote cooperation, highlighting the crucial role played by both the initial placement of cooperators and the presence of hyperdegree correlations. Overall, our results provide a comprehensive characterization of nonlinear PGGs on hypergraphs and open up new avenues for richer models of evolutionary dynamics of multiplayer interactions on structured populations.

DNA language models like Evo2 now fit million-token contexts large enough to cover entire TADs, yet whether they learn 3D chromatin structure, a key regulatory layer acting atop primary sequence, remains untested and questionable, given that Evo2's training data includes prokaryotes lacking this structure. We probed Evo2-7B on TAD boundaries and convergent CTCF loops in 1 Mb windows using two complementary tests: likelihood-based perturbation and sequence generation. Evo2 did not distinguish functional perturbations from matched random controls and failed to reliably generate convergent CTCF loops, recovering TAD boundaries only partially. Together, these results indicate that Evo2 has learned local CTCF grammar but misses higher-order 3D organization, pointing to bidirectional model architectures integrating cell types and 3D contacts, rather than longer contexts, as the path to developing 3D-aware DNA language models.

Systems of differential equations have been used to model biological systems such as gene and neural networks. A problem of particular interest is to understand the number of stable steady states. Here we propose conjunctive networks (systems of differential equations equations created using AND gates) to achieve any desired number of stable steady states. Our approach uses combinatorial tools to predict the number of stable steady states from the structure of the wiring diagram. Furthermore, AND gates have been successfully engineered by experimentalists for gene networks, so our results provide a modular approach to design gene networks that achieve arbitrary number of phenotypes.

Motivation: Protein folding is a dynamic process during which a protein's amino acid sequence undergoes a series of 3-dimensional (3D) conformational changes en route to reaching a native 3D structure; the resulting 3D structural conformations are called folding intermediates. While data on native 3D structures are abundant, data on 3D structures of non-native intermediates remain sparse, due to limitations of current technologies for experimental determination of 3D structures. Yet, analyzing folding intermediates is crucial for understanding folding dynamics and misfolding-related diseases. Hence, we search the literature for available (experimentally and computationally obtained) 3D structural data on folding intermediates, organizing the data in a centralized resource. Additionally, we assess whether existing methods, designed for predicting native structures, can also be utilized to predict structures of non-native intermediates. Results: Our literature search reveals six studies that provide 3D structural data on folding intermediates (two for post-translational and four for co-translational folding), each focused on a single protein, with 2-4 intermediates. Our assessment shows that an established method for predicting native structures, AlphaFold2, does not perform well for non-native intermediates in the context of co-translational folding; a recent study on post-translational folding concluded the same for even more existing methods. Yet, we identify in the literature recent pioneering methods designed explicitly to predict 3D structures of folding intermediates by incorporating intrinsic biophysical characteristics of folding dynamics, which show promise. This study assesses the current landscape and future directions of the field of 3D structural analysis of protein folding dynamics.

A key unresolved question in microbial ecology is how the extraordinary diversity of microbiomes emerges from the behaviour of individual populations. This process is driven by the cross-feeding networks that structure these communities, but are hard to untangle due to their inherent complexity. We address this problem using the tools of network science to develop a model of microbial community structure. We discover tipping points at which diversity abruptly declines due to the catastrophic collapse of cross-feeding networks. Our results are a rare example of an ecological tipping point in diversity and provide insight into the fundamental processes shaping microbiota and their robustness. We illustrate this by showing how the unculturability of microbial diversity emerges as an inherent property of their microbial cross-feeding networks.

Recent work has addressed the problem of inferring Langevin dynamics from data. In this work, we address the problem of relating terms in the Langevin equation to statistical properties, such as moments of the probability density function and of the probability current density, as well as covariance functions. We first review the case of linear Gaussian dynamics, and then consider extensions beyond this simple case. We address the question of quantitative significance of effects. We also analyze underdamped (second-order) processes, specifically in the limit where dynamics in state space is almost Markovian. Finally, we address detection of non-Markovianity.

We present the first systematic study of when target context helps molecular property prediction, evaluating context conditioning across 10 diverse protein families, 4 fusion architectures, data regimes spanning 67-9,409 training compounds, and both temporal and random evaluation splits. Using NestDrug, a FiLM-based architecture that conditions molecular representations on target identity, we characterize both success and failure modes with three principal findings. First, fusion architecture dominates: FiLM outperforms concatenation by 24.2 percentage points and additive conditioning by 8.6 pp; how you incorporate context matters more than whether you include it. Second, context enables otherwise impossible predictions: on data-scarce CYP3A4 (67 training compounds), multi-task transfer achieves 0.686 AUC where per-target Random Forest collapses to 0.238. Third, context can systematically hurt: distribution mismatch causes 10.2 pp degradation on BACE1; few-shot adaptation consistently underperforms zero-shot. Beyond methodology, we expose fundamental flaws in standard benchmarking: 1-nearest-neighbor Tanimoto achieves 0.991 AUC on DUD-E without any learning, and 50% of actives leak from training data, rendering absolute performance metrics meaningless. Our temporal split evaluation (train up to 2020, test 2021-2024) achieves stable 0.843 AUC with no degradation, providing the first rigorous evidence that context-conditional molecular representations generalize to future chemical space.