COVID-19 and previous pandemics have shown how diseases can disrupt, threaten, and transform daily life. Since pathogens and societies are continuously evolving, every pandemic is different. However, certain fundamental principles of disease transmission appear to hold true across different outbreaks. These ``mechanisms'' are grounded in natural laws or the very structure of our biology and societies. This paper compiles ten fundamental mechanisms, curated by a multidisciplinary team with backgrounds spanning public health, medicine, epidemiology, political science, mathematics, physics, and psychology. These mechanisms, although perhaps underappreciated, substantially shape how pandemics unfold and are controlled. The better we succeed in understanding these mechanisms and establishing this knowledge in our societies, the better we will be able to prepare for future pandemics and respond appropriately when they occur.

Across biological and social systems, cooperation often depends on phenotypic cues rather than random encounters. To account for real-world interactions unfolding across multiple, simultaneous dimensions, here we develop a general framework for the evolution of cooperation in multiplex networks governed by multi-phenotype homophily. We derive analytical conditions for natural selection to favor cooperation across phenotypic traits that are independent or exhibit epistasis and under different modes of mutation coupling. Despite the integration of fitness across layers, the conditions for cooperation resolve into layer-specific $\sigma$-rules, depending only on the local payoff structure, the effective number of phenotypes, and the mutation rates. We show that phenotypic diversity fosters cooperation by partitioning populations into assortative niches. Furthermore, in finite populations, intensifying the prisoner's dilemma shifts the dependence of cooperation on strategy mutation from monotonically decreasing, through U-shaped, to monotonically increasing. Our work provides a unified account of how multi-phenotype homophily underpins the evolutionary dynamics of cooperation in heterogeneous populations.

Science is constitutively third-personal: its findings are in principle reproducible by any observer, independent of perspective, and answerable to measurement. This is the source of its power and also its limit when it comes to phenomena that are first-personal. While it is obvious that a science of the Meaning of Life is unattainable, researchers have not drawn the same conclusion for consciousness -- in its phenomenal dimension, the qualia of seeing red, of feeling pain, of being anything at all. I argue they should. The hard problem of consciousness is not a scientific problem awaiting better tools or a more ambitious theory, but a category error. The same structural problem applies to machine consciousness: neither attribution nor denial is scientifically adjudicable. I situate science within a broader ecology of understanding and argue that a unified framework that addresses both the objective and the subjective may be unattainable.

The spectral behaviour of the synaptic matrix, representing the neuronal connection strengths, is an important tool to analyze the stability and transient dynamics of a typical brain as well as its learning process and memory capacity. The complexity of the brain due to large number of neurons as well as underlying transient mechanisms e.g. homeostasis, seizure or synaptic plasticity can lead to networks with time-varying degree and type of sparsity. This renders an exact determination of the synaptic matrix not only technically difficult but also meaningless, leaving its statistical analysis as the best available theoretical approach. This motivates us to pursue a spectral analysis of the synaptic matrix models with different type of sparsity and thereby analyze latter's role on various aspects of network dynamics and stability. Our results have potential relevance for detemining the type of synaptic sparsity required to induce a specific brain function or desired transient mechanism e.g for pharmacological effects or physiological modulators.

Metastable states, a phenomenon observed in brain dynamics and many other systems, have been proposed as a key feature of healthy brain function, reflecting a balance between integration and segregation. However, it remains unclear how to capture this behavior within a dynamical-systems framework. In this paper, we propose sequential chaotic oscillations (SCOs), arising in excitatory-inhibitory threshold-linear networks (E-I TLNs), as a candidate dynamical mechanism for sequential metastability. As a simple form of chaotic itinerancy, SCOs occur under constant input and consist of a sequence of metastable states whose transition order can be predicted by the underlying graph. To identify the parameter regime for SCOs, we develop new graph rules for E-I TLNs and use them to characterize the fixed point structure of E-I TLNs on paths and cycles. Our results show that the emergence of SCOs requires unstable singleton fixed points and sufficiently strong inhibition. In addition to SCOs, we find that E-I oscillations need not be synchronized. Motivated by this, we introduce a decomposition into the z-mode and the mean mode, which capture excitatory differences and overall network activity, respectively. These modes are then used to distinguish attractors associated with the full-support fixed point of E-I TLNs on cycles.

Genotype-based cis-expression prediction depends on accurately modeling local regulatory architecture. We present block-sparse Bayesian sparse linear mixed model (bsBSLMM), an extension of Bayesian sparse linear mixed model (BSLMM) that incorporates linkage disequilibrium (LD)-block spike-and-slab sparsity and a transcription start site (TSS)-informed SNP inclusion prior. Across 23,098 genes from GEUVADIS European-ancestry lymphoblastoid cell lines, bsBSLMM retained more predictable genes than BSLMM, LASSO, BLUP, TIGAR elastic net, and TIGAR Dirichlet-process regression under matched evaluation criteria. Compared with BSLMM, bsBSLMM improved held-out prediction performance for most shared genes, with gains driven primarily by LD-block sparsity and further enhanced by the TSS-informed prior. Variants selected by bsBSLMM showed stronger enrichment in GM12878 DNase and H3K27ac regulatory regions than variants selected by BSLMM. In transcriptome-wide association study (TWAS) analysis, bsBSLMM recovered established inflammatory bowel disease signals, including IL23R, and identified additional genome-wide significant genes not detected by BSLMM. Independent validation in the Louisiana Osteoporosis Study reproduced the increased prediction yield across ancestries and recovered biologically relevant bone mineral density pathways in downstream TWAS and gene set enrichment analyses. These results demonstrate that incorporating LD-block structure and biologically informed SNP priors improves cis-expression prediction and enhances downstream TWAS discovery.

It has been proposed that the brain integrates flexible, computationally expensive cortical processing with simpler, lower-cost subcortical mechanisms to achieve resource-efficient performance greater than that of either system alone. Despite the allure of this perspective, satisfying theoretical frameworks that explore this hypothesis are still limited. We extend existing frameworks in which a model-based module and model-free module learn in tandem by explicitly constraining the memory resources of the model-based module, and investigate the impact of this constraint in a simple decision-making setting. Memory constraints naturally give rise to strategies for allocating memory resources. We evaluate the performance of different strategies in different situations and demonstrate that when the rewarded states change often, it can be advantageous for the model-based module to focus its memory resources not on exploiting the current reward, but on capturing general structure of the environment. This work provides a theoretical foundation for a functional dissociation between cortical and subcortical systems during learning: the cortex supports general structure learning, while subcortical circuits specialize in reward-based learning. We further detail how these hypotheses can be tested on experimental data.

We present a multimodal dataset of 1020 hours of simultaneously recorded scalp electroencephalography (EEG), facial electromyography (EMG), and speech audio from three healthy native Japanese speakers during open-vocabulary overt speech. Recordings were acquired with three EEG systems-an ultra-high-density system (this http URL) and two cap-type systems (this http URL and eegosports), spanning 62-128 channels-across many sessions over several months. Each session provides time-synchronized EEG, facial EMG, and audio, together with speech-event annotations and transcriptions. Although collected with speech decoding as a primary motivation, the dataset also supports work on multimodal signal processing, artifact modeling, longitudinal and cross-device adaptation, and EEG representation learning. Technical validation included power spectral density and event-related potential analyses across participants, devices, and tasks, which showed the expected 1/f spectral profile, task-related alpha-band attenuation, and time-locked evoked responses. The dataset is released in Brain Imaging Data Structure (BIDS) format via OpenNeuro under a CC0 waiver to support both speech-related and broader EEG research.

Hypergraphs provide a natural framework for representing neurophysiological interactions distributed across sets of sensors. A key methodological question is how hyperedges should be defined from frequency-resolved electroencephalography/magnetoencephalography (EEG/MEG) data. We demonstrate a construction strategy in which hyperedges are obtained from canonical coherence (caCoh), an extension of coherence that estimates coupling between multidimensional signal spaces. To our knowledge, this is the first work to construct hypergraphs directly from a multivariate connectivity measure specifically designed for frequency-resolved neurophysiological analysis. We propose two caCoh-based representations: a one-to-space hypergraph, where each external signal defines a hyperedge over the EEG/MEG sensor space, and a space-to-space hypergraph, where two multidimensional signal spaces are represented by a single hyperedge. We evaluate the approach in controlled simulations with known coupling frequencies and varying signal-to-noise ratio (SNR). Compared with graphs based on magnitude-squared coherence (MSC), caCoh-based hypergraphs showed statistically higher target-baseline contrasts at almost all SNR levels, indicating stronger recovery of coupling frequencies. They also recovered sensor-level spatial patterns associated with the simulated sources. In addition, one-to-space and space-to-space representations reduced 610 MSC edges per frequency to 10 and 1 hyperedges, respectively. These results establish multivariate spectral connectivity as a natural methodological basis for EEG/MEG hypergraphs.

In this study, we predicted the structure of the heptapeptide APRLRFY, a neuropeptide sequence, on a tetrahedral lattice using a Quantum Approximate Optimization Algorithm (QAOA). QAOA is based on the adiabatic approximation and has been successfully applied to a wide range of optimization problems. However, relatively slow convergence during ground-state searches has frequently been reported. To overcome this limitation, we employed the Counter-Diabatic Quantum Approximate Optimization Algorithm (CD-QAOA), which introduces an additional counter-diabatic driving term into the adiabatic framework to accelerate convergence toward the ground state during peptide structure prediction. In the heptapeptide structure prediction, intermolecular interactions were modeled using two different approaches. In the first approach, only the interaction between the second residue, proline (P), and the seventh residue, tyrosine (Y), was included in the optimization. In the second approach, all residue-residue interactions within the heptapeptide were modeled using the Miyazawa-Jernigan (MJ) interaction matrix. To validate the peptide structures predicted using CD-QAOA, we additionally employed several classical computational methods, including quantum chemistry-based Hartree-Fock (HF) calculation and Density Functional Theory (DFT) calculation, conventional molecular dynamics (MD) simulation, and Hamiltonian replica exchange molecular dynamics (H-REMD) simulation. The structural similarities among the conformations obtained from these different approaches were systematically analyzed. CD-QAOA is highly effective for predicting the structures of short peptides. In particular, we demonstrate that a quantum-classical hybrid framework can significantly improve both the efficiency and accuracy of peptide structure prediction.

Biomolecules such as proteins and small-molecule ligands play a central role in biological systems, arising from the tight interplay between sequence and three-dimensional structure. Recent generative models for biomolecular co-design aim to capture this interplay by jointly modeling coupled modalities. However, existing approaches largely adopt a parallel execution of marginal generative processes, implicitly enforcing fixed synchronous coupling. We argue that a critical but overlooked degree of freedom lies in how these marginal processes are temporally coupled during training and generation, where inappropriate coupling can introduce high-variance supervision and inconsistent intermediate states, affecting modality consistency. To address this, we introduce GeoCoupling, a systematic framework that optimizes for temporal couplings between heterogeneous modalities. Empirical results across structure-based drug design and unconditional protein design demonstrate the learned couplings consistently outperform synchronous and randomly coupled baselines, yielding biomolecules with improved physical validity and diversity.

The outer membrane is the interface through which Gram-negative bacteria - a broad classification of organisms including \textit{Escherichia coli} and a number of deadly pathogens - interact with the environment. Two decades of work on the process of outer membrane biogenesis have led to the discovery of the components that mediate this process, and the characterisation of structure and function of these component parts of the bacterial cell machinery. However, neither current experimental methods, nor conventional molecular dynamics (MD) simulation approaches are capable of investigating this membrane machinery on the time scale of the cell division cycle. This leaves crucial questions unanswered, such as how this lipid-poor, largely static environment is organised to permit ongoing membrane growth. Here, we introduce a semi-quantitative agent-based model to explore the molecular-scale dynamics of Gram-negative outer membrane as it grows. Model simulations across a broad region of parameter space suggest that protein incorporation into the membrane by the $\beta$-barrel assembly machinery (BAM complex) is a process which is prone to stalling, and may take place only in short bursts. We also find suggestions that BAM complexes work collaboratively with each other, and with the lipopolysaccharide-inserting Lpt complex when in close proximity. The agent-based framework we introduce provides a means to assess and generate hypotheses on outer membrane biogenesis on previously inaccessible time scales.

Biological neurons receive thousands of synaptic inputs on branching, electrically excitable dendrites, yet population activity is often modeled with direct input-output rules in which each input contributes independently to a scalar drive. We study what successful prediction by such models does, and does not, reveal about neural computation. For conditional maximum-entropy models that match output rates and pairwise output-input coactivities, the entropy explained by a direct model is a prediction measure under the sampled input distribution, not a mechanism-identification test. A restricted MaxEnt fit is an information projection: omitted interaction, temporal, or hidden-state terms can be absorbed into fitted first-order parameters whenever they are correlated with the included sufficient statistics. For sparse correlated binary inputs, this absorption has an explicit coskewness form. We introduce diagnostics that separate in-distribution prediction from recovery of the response rule: state reweighting that holds P(y|x) fixed while changing P(x), conditional log-odds contrasts for local additivity, and temporal leakage controls. In ground-truth simulations, purely higher-order responses can pass first-order entropy and raw coactivity tests under leakage-prone sampling, but are correctly classified after reweighting. Applied to selected, leakage-enriched local tables from CA1 hippocampal recordings, approximately half of tables that appear first-order under empirical weights become distribution-sensitive under balanced reweighting, far above a matched additive-surrogate null. Thus direct entropy-explained fractions and raw coactivity predictions should be interpreted as predictions under the observed state distribution, not as evidence that mechanisms outside the direct model are absent or small.

Selecting where to intervene on a protein (i.e., choosing a targetable site) is often a more ambiguous and failure-prone bottleneck than selecting what binds, especially for membrane proteins where accessibility, topology, and post-translational modifications (PTMs) constrain actionable regions. We present Site4Drug, a modality-aware site-finding agent that outputs a ranked list of targetable regions with explicit constraints, evidence summaries, risk flags, and a traceable decision log. Rather than requiring users to specify the drug modality upfront, Site4Drug can recommend a binding modality (e.g., antibody/peptide-like vs small-molecule) from the same evidence used for site discovery, including topology, hydropathy, PTM propensity, disulfides, domain context, and sequence. Importantly, this evidence is applied consistently across modalities, including small-molecule pocket discovery, to avoid selecting chemically plausible but biologically occluded sites.

Live neural systems demonstrate remarkable capabilities to learn new behavior and patterns from mere few examples and are known to operate robustly under severe sensory noise. These capabilities, however, remain largely out of reach for modern artificial neural networks, including deep learning models. We show that this gap can be bridged by embedding novel genuine neuromorphic circuits into conventional artificial neural network architectures. These circuits comprise astrocytic modulation and spiking dynamics inherent to biological neural structures. Tested across standard benchmarks representing tasks of varying complexity, the hybrid models achieve high accuracy from few training examples per class and sustain high performance under occlusion and impulse noise that cause performance collapse in standard models without neuromorphic adaptation. We term this phenomenon neuromorphic supremacy - a regime in which architectures grounded in neurobiology decisively outperform classical deep learning, pointing toward a principled foundation for perception in embodied AI systems operating in noisy, data-scarce environments.

INTRODUCTION. This study presents the first quantification of vertex-level grey-matter associations between Alzheimer's disease (AD) and Parkinson's disease (PD) using highresolution brain maps aggregated from large MRI datasets. The aim is to identify shared neuroanatomical signatures between the two diseases. METHODS. Leveraging a novel statistical framework (SumR2 regression), adapted from genetic correlation analysis, we estimated the shared neuroanatomical signature (grey-matter correlation: rGM) between AD and PD. RESULTS. A significant positive brain-wide grey-matter correlation (rGM=0.24, 95%CI 0.20-0.28) was observed between AD and PD. This correlation was further observed across disease stages and replicated using UK Biobank data. We located 9 vertex-wise clusters (106 vertices) that contribute to the significant rGM, highlighting reduced thickness in the bilateral putamen and right accumbens as associated with both AD and PD. DISCUSSION. Our findings suggest that shared neuroanatomical features emerge early in neurodegeneration and have implications for early screening, disease monitoring, and targeted interventions. from the Parkinson's Progression Markers Initiative (PPMI) database (this http URL

Progress in AI is turning machine consciousness from a philosophical curiosity into a societal issue, and has led to criticism of the widespread computational functionalism framework. Biological Naturalism (BN) claims that biology, not computation, is crucial for consciousness. We discuss which forms of BN are empirically testable. For Type-A-BN, biology intrinsically matters for consciousness, without affording unique information processing capabilities. We argue, similarly to the unfolding argument, that this dissociates consciousness from behaviour, making Type-A-BN untestable. For Type-B-BN, biology matters because it affords unique information processing capabilities. Type-B-BN is testable, and not incompatible with computational functionalism. Both face the same task: relating consciousness to information processing. Biology can act as a guide on this quest, but not as a solution.

Understanding how speech foundation models relate to human cortical activity is a key challenge for computational neuroscience. Here, we investigate how internal representations from Whisper predict intracranial ECoG responses during naturalistic speech perception. We introduce a time-resolved neural encoder that combines speech embeddings with a recurrent temporal model and soft attention, allowing us to examine layer-wise brain alignment. Intermediate Whisper layers provide the strongest correspondence with neural activity, supporting a hierarchical match between model representations and cortical speech processing. Comparisons with baselines show that high-resolution ECoG responses benefit from temporally structured modelling beyond linear mappings from the same speech representations. In addition, attention maps reveal temporally local alignment between speech embeddings and neural responses, while a phonemic interpretability analysis identifies anatomically coherent phoneme-category organization among encoding-informative electrodes. Together, these results suggest that speech foundation models offer a useful framework for studying time-resolved cortical speech representations.

Sparse Autoencoders (SAEs) have found success parsing neural representations into interpretable concepts, providing a basis for understanding and control. However, what exactly SAEs extract, and, correspondingly, the scientific conclusions we can draw from them, are not obvious. Empirically, the proof is in the pudding: SAEs learn interpretable features. Theoretically, we lack a clear account of what properties a 'concept' must satisfy for an SAE to extract it. There has been extensive identifiability work studying the conditions under which sparse coding recovers ground-truth features; however, these approaches tends to focus on simple data-generating models (e.g. sparse independent features) which poorly approximate the internet-swallowing language-model representations on which SAEs are trained. Here, avoiding data-generating models, we ask simply what properties any dictionary learning optimum must satisfy. Concretely, we extend local optimality analyses (Gribonval & Schnass, 2010) to the nonnegative joint-optimisation problem that vanilla SAEs approximate, and derive constraints relating optimal SAE features to their distributions. We use these constraints to explain a range of observed SAE behaviours - hierarchical splitting & absorption, the structure of residuals, and dense antipodal features - each reflecting how L1+nonnegativity interact with data to structure optimal dictionaries. Finally, we construct a novel large-dictionary convex problem and explore the wide atom-per-datapoint limit. In sum, we hope to tease model assumptions from unexpected observations, letting us learn more from SAEs' successes and provide principles for designing their successors.

Computational epitope prediction is a critical tool for exploring and understanding CD4+ T cell-mediated immune responses, a key aspect of adaptive immunity. While existing computational methods primarily focus on supervised learning approaches, they often overlook the essential role of antigen processing in determining binding specificity. To address this limitation, our group developed Antigen Processing Likelihood (APL), an algorithm that integrates crystallographic B-factor, solvent accessible surface area (SASA), hydrogen exchange protection factors (COREX), and sequence entropy. In this paper we introduce APLSuite, a comprehensive and lightweight software suite designed to streamline APL-based epitope prediction. APLSuite integrates distributed RESTful API services, a Python client for data aggregation and processing, a data science tool for efficient epitope computation, and a user-friendly graphical user interface for non-coding users. It provides a seamless and efficient pipeline for APL calculation and epitope prediction that can be finished in minutes with GPU-acceleration, which has not been implemented by existed tools. This flexible and extensible software suite is deployable on desktop and cloud environments, offering both guided and customizable workflows to meet diverse research needs in immunology research and immunotherapy development.

Biological and neuromorphic recurrent neural networks (RNNs) are subject to spatial and temporal locality constraints on the information that can plausibly be used during learning. A common strategy to satisfy these constraints is to modify gradient descent by neglecting non-local terms to varying degrees, as in random feedback local online (RFLO) learning and truncated backpropagation through time (tBPTT). However, the learning dynamics of these algorithms, and how they compare with BPTT, remain poorly understood. We apply dynamical systems theory to data-aligned linear RNNs -- whose dynamics can be separated into orthogonal modes -- to compare stationary solutions, stability properties, and convergence rates, finding qualitatively distinct behaviour for RFLO versus BPTT and one-step tBPTT. We further observe that the solutions learned by RFLO are restricted to low-rank perturbations of initial parameters, a result which holds beyond the data-aligned setting. Our work provides analytical insight into how locality constraints shape learning dynamics, with implications for neuroscientific models of learning and alternative optimization approaches for RNNs.

Quenched disorder in the environment can fundamentally alter transport dynamics in both active and passive systems. We explore how disordered arrays of filaments govern the distribution of intermittently moving particles which switch between diffusive and processive transport. Motivated by the mixed-polarity arrangements of parallel microtubules observed in mammalian dendrites, we show that such arrays tend to result in localization of particles at regions of convergent filament orientation. In the rapid attachment-detachment limit, the disordered system can be described by a noisy one-dimensional effective energy landscape, whose structure is approximated by a random walk. The depth and width of wells on this landscape are expressed as a function of the transport kinetics and system geometry. Localization is shown to be strongest at intermediate run-lengths, where biased transport persists long enough to sense the quenched filament polarity but not so long as to facilitate escape from local traps. These results demonstrate robust localization of particles moving on random filament networks, highlighting the emergent spatial organization that arises from an interplay of active transport and quenched disorder.

Structure-based drug design increasingly employs LLM agents to iteratively refine ligands against a target pocket, yet a viable ligand must satisfy two often-conflicting objectives -- binding affinity and druggability -- which single optimization steps rarely improve together. To quantify this difficulty, we introduce two diagnostic metrics: the first measures how often a single edit improves both objectives, and the second measures how often a gain on one objective comes with a loss on the other. Applying these diagnostics to current LLM-agent pipelines exposes a consistent failure mode: the agent performs molecular editing without knowing how the pocket-ligand complex responds to local modifications, thus rarely achieving joint improvement. Inspired by medicinal chemists, who probe the pocket-ligand complex with controlled analog edits before choosing an optimization direction, we propose \textbf{PROBE}, an optimization framework built around edit-response probing. PROBE first decomposes the ligand into editable sites and builds a pocket-specific \textbf{site map} that flags where joint gains are plausible, where the two objectives are likely in tension, and where liability substructures should be changed; it then performs controlled probe edits whose responses are distilled into an \textbf{EditManual}. Guided by the site map and EditManual, PROBE runs an iterative multi-agent loop in which an affinity agent, a druggability agent, and a co-optimization agent jointly produce edits. On the CrossDocked2020 benchmark, PROBE achieves state-of-the-art performance and substantially mitigates the failure modes exposed by our diagnostics metrics.

Bulk gene expression profiling, which aggregates pooled RNA across cells within a biological sample, remains important in the single-cell era because it is typically less noisy, more sensitive, and more cost-effective than single-cell assays. Accordingly, a growing body of computational methods seeks to recover causal relations among genes from bulk expression data. However, aggregation is a lossy, non-invertible coarsening of the underlying cellular system, and it remains unclear whether and under what conditions causal relations are recoverable from aggregated bulk gene expression data. To answer this, we formalize recoverability under aggregation through two notions of consistency: functional-form consistency and conditional-independence consistency. We then derive necessary and sufficient conditions for recoverability, showing that these properties are preserved only under linear aggregations (e.g., sum/mean) coupled with affine structural equations. To assess the practical plausibility of these conditions, analyses of four bulk and four single-cell gene expression datasets further reveal that the estimated pairwise regulatory functions among genes deviate from linearity in both data types, providing limited empirical support for the linearity assumptions required for recoverability. Together, these results caution against recovering causal relations from aggregated bulk expression data without strong additional assumptions.

Despite the growing availability of cryo-electron microscopy (cryo-EM) density maps, effectively leveraging them for protein representation remains challenging. First, current methods lack a general-purpose protein pretraining framework tailored for cryo-EM density maps, designed for protein-related property prediction. Second, existing approaches typically partition density maps into local box regions and model them independently, overlooking interactions across boxes which are essential for capturing global structural context in cryo-EM density map. To address these challenges, we propose CryoProt, a protein pretraining framework designed for cryo-EM density maps. CryoProt introduces a Map Encoder based on multi-head latent attention (MLA), where box-level representations interact through a shared latent space, enabling explicit modeling of cross-box dependencies within the density map. Furthermore, we adopt a multi-task pretraining strategy to learn generalizable representations that can be effectively transferred to diverse downstream tasks, such as protein flexibility prediction, where cryo-EM density maps are not required and can be inferred implicitly by the pretrained model. Experimental results demonstrate that CryoProt consistently outperforms existing state-of-the-art methods across multiple benchmarks, achieving up to 12% improvement over the best-performing baselines, highlighting the importance of modeling cross-box interactions in cryo-EM data. The source code is publicly available at this https URL.

Biochemical systems involve both the flow of matter, in which entities transform into one another via reactions, and the flow of information, in which entities regulate which reactions may occur. Boolean networks capture the latter; reaction networks capture the former. Yet no unified qualitative formalism treats regulated reactions as its principal objects of study, despite their prominence in standards such as the Systems Biology Graphical Notation Process Description (SBGN-PD) language. We introduce modulation-reaction networks (MR-networks), a mathematical framework in which entities modulate reactions through activations and inhibitions, and study their synchronous Boolean semantics. To reason about MR-networks we develop Modulation-Reaction Logic (MRL), a hybrid modal $\mu$-calculus whose modalities reason about the structure of the network and whose fixed-point operators capture temporal evolution of the computation. We establish a collection of validities, including a complete characterisation of the one-step update rule, and demonstrate the expressive power of MRL by formalising properties of biological interest such as reachability, sustained production, and presence of attractors. We show that MRL admits model-checking via an evaluation game, and introduce a bisimulation relation for MR-networks, which is proved to be invariant for all MRL-formulas. As a step towards a biologically more realistic computational model, we sketch the asynchronous semantics of MR-networks, and outline how the developments for the synchronous case transfer to the study of the asynchronous one.

Many networks not only support but also rely on transient non-normal amplification, an orders-of-magnitude increase in the activity of an otherwise stable system. Constructing such networks under hard sign/sparsity/diagonal constraints -- the regime relevant for biological connectomes and structured RNN initializations -- has so far required either gradient-based local search with thousands of inner-loop eigendecompositions or Schur-form direct construction in an abstract basis that breaks the constraints under projection. Here we introduce DAGGER (Directed Acyclic Graph Guided Edge Reweighting), a gradient-free single-pass algorithm. Given a stable signed sparse matrix, DAGGER produces an output with the same sign, sparsity, and diagonal. A single scalar $\beta$ controls a Wasserstein-2 budget that smoothly trades exact multiset preservation ($\beta = 0$) for amplification; peak amplification grows essentially without bound with $\beta$, empirically reaching $10^{10}$ before numerical overflow. DAGGER matches or exceeds gradient-based methods at multiset preservation in a single forward pass -- 30-100$\times$ fewer eigendecompositions than a typical gradient inner loop -- and at moderate $\beta$ beats them by orders of magnitude with connectivity exactly preserved. We develop the algorithm, compare it to the existing methods and on a downstream signal-detection task, and examine the diagnostics that show why DAGGER is structurally different from other amplifying networks.

We show that computing the log-partition function (free-energy) of conditioned inhomogeneous Curie--Weiss spin Hamiltonians reduces to an unbalanced $2 \to 1$ norm computation, and design a polynomial-time SDP algorithm for this problem with a lower bound proof for the amount of unbalance achieved. Applied to the protein Ubiquitin, the framework starts from a known crystal structure, explores alternative backbone conformations across the free-energy landscape, and identifies flexible regions of the protein while preserving its native secondary structure.

Protein language models (PLMs) are passive oracles: they generate sequences in a single forward pass with no mechanism to consult external biophysical feedback or redirect generation when a candidate violates thermodynamic or structural constraints. We introduce AgentPLM, which addresses this by equipping a pre-trained PLM with i) Reasoning-Augmented Decoding (RAD), which interleaves autoregressive generation with tool calls (ESMFold, FoldX, AutoDock Vina), and ii) Contrastive Agent Policy Optimisation (CAPO), a trajectory-level extension of direct preference optimisation that trains the policy end-to-end to learn when oracle feedback is informative rather than merely imitating high-fitness sequences. We evaluate AgentPLM on benchmark tasks spanning de novo enzyme design, antibody optimisation, thermostability, PPI interface design, and zero-shot fitness prediction with standardised oracle APIs and controlled sequence-identity splits. AgentPLM achieves state-of-the-art results with a gain in antibody top-10% hit rate over the strongest passive baseline, providing mechanistic evidence of online error correction without explicit backtracking.

We ask whether dependency topology correlates with functional load-bearing organization as recoverable geometry -- not as a metaphor, but as a measurable structural property detectable by multilayer network analysis. Across seven independent substrates, we show that hub persistence and rank divergence under the Functional Proximity Law recover operational organization that domain experts describe as logic: axiomatic load-bearing structure in formal mathematics, control and contract structure in legacy software, conserved hub grammar across approx. 600 million years of neural evolution, catalytic role organization in a published prebiotic autocatalytic network, carry-path dominance in a 4-bit digital circuit, betweenness persistence in the ISCAS85 c432 standard benchmark (n=196), and a directional formal-systems replication in the Coq Corelib (n=17). A key methodological finding: degree-based hub persistence is weak between physical wiring and simulation state-correlation layers (r=0.21 in c432), while betweenness-based persistence is stronger (r=0.77 in the 4-bit ALU post-hoc; r=0.34 in c432). The ISCAS85 pre-registered primary hypothesis was CONFIRMED (degree r=0.426, p=0.002, Spearman r=0.551). The formal-systems claim is supported by two proof-assistant corpora: Lean 4 mathlib4 (CONFIRMED, r=0.777, p=0.004) and Coq Corelib (PARTIAL, direction confirmed, r=0.288, p=0.287, n=17, underpowered). All seven experiments were pre-registered before analysis.

We formulate a structure-informed multiple sequence alignment problem, denoted MSA-S. The model abstracts biological sequences as strings and structural information as designated position-pairs. It augments a fixed pairwise string score, defined by a fixed non-gap symbol-pair scoring rule and fixed affine gap penalties, with a binary overlap score on designated position-pairs, which can be interpreted as a contact-map overlap score in structural applications. This yields a fixed-score, integer-valued optimization model suitable for complexity-theoretic analysis. Under this formulation, we show that the decision problem MSA-S-DEC is NP-complete for a broad class of fixed pairwise string scoring schemes. We also show that NP-hardness persists even under the restriction that every designated position-pair set is nonempty and the pair-overlap threshold is strictly positive. For the associated scalarized optimization problem MSA-S-OPT(lambda) with any fixed rational constant lambda >= 1, we further show that, under the canonical unit scheme for the non-gap symbol-pair scoring rule, MSA-S-OPT(lambda) admits no polynomial-time approximation scheme (PTAS) even for two input strings (k = 2), unless P = NP. These results establish a formal complexity-theoretic baseline for structure-informed multiple sequence alignment.

A cholera transmission model is formulated that incorporates water-borne and horizontal transmissions as well as infectivity of deceased individuals. The model includes an Allee effect for the bacteria in the environment and imperfect and waning vaccination. Mathematical properties of the model are investigated, with an environmental bistability shown to combine with a vaccine-driven one, although a computational search for the latter fails to detect its presence in realistic parameter ranges. The computational analysis also considers the interplay between vaccination strategy, vaccine efficacy and waning, as well as the effect of transmission of the disease during funeral rites. The effect of control scenarios such as WASH or Safe and dignified burials are assessed.

Respiratory viral infections pose a global health burden, yet the cellular immune mechanisms underlying protection and pathology remain unclear. Natural infection cohorts often lack pre-exposure baselines and time-controlled sampling, whereas inoculation and vaccination trials generate well-structured longitudinal transcriptomic data. However, these datasets are scattered across repositories and processed inconsistently, hindering integrative and AI-driven analyses. To address these challenges, we developed the Human Respiratory Viral Immunization LongitudinAl Gene Expression (HR-VILAGE-3K3M) repository: an AI-ready resource integrating bulk and single-cell transcriptomic profiles from 3,178 subjects across 66 studies. The dataset spans vaccination, inoculation, and mixed exposures, with samples from blood and nasal swabs collected from public repositories including GEO, ImmPort, and ArrayExpress. We curated and harmonized subject-level metadata, standardized outcome measures, and applied unified preprocessing with rigorous quality control. We further provide benchmark analyses illustrating its utility. This resource supports discovery of biomarkers, immune mechanisms, and methodological development. As one of the largest longitudinal transcriptomic resources for human respiratory viral immunization, HR-VILAGE-3K3M enables reproducible and scalable analyses to accelerate vaccine and antiviral research.

Selecting an effective docking algorithm is highly context-dependent, and no single method performs reliably across structural, chemical, and protocol regimes. MolAS is a lightweight algorithm-selection model that predicts per-algorithm performance from pretrained protein and ligand embeddings using attentional pooling and a shallow residual decoder. With hundreds to a few thousand labelled complexes, MolAS achieves up to a 15 percentage-point absolute improvement over the single-best solver (SBS) and closes 17--66\% of the Virtual Best Solver (VBS)--SBS gap across five docking benchmarks. Analyses of selection frequencies, margin-conditioned reliability, and benchmark-level oracle structure indicate that MolAS is most effective when the workflow-defined oracle landscape has low winner entropy and a reasonably separable top-solver region, but degrades under protocol mismatch that shifts solver rankings and changes the induced labels. These results suggest that, in the evaluated regime, robustness is limited less by representational capacity than by workflow- and protocol-induced instability in solver hierarchies, positioning MolAS as an in-domain selector for fixed pipelines and as a diagnostic tool for assessing when docking algorithm selection is well-posed.

The study of cortical dynamics during different states such as decision making, sleep and movement, is an important topic in Neuroscience. Modelling efforts aim to relate the neural rhythms present in cortical recordings to the underlying dynamics responsible for their emergence. We present an effort to characterize the neural activity from the cortex of a mouse during natural sleep, captured through local field potential measurements. Our approach relies on using a discretized Wilson--Cowan Amari neural field model for neural activity, along with a data assimilation method that allows the Bayesian joint estimation of the state and parameters. We demonstrate the feasibility of our approach on synthetic measurements before applying it to a dataset available in literature. Our findings suggest the potential of our approach to characterize the stimulus received by the cortex from other brain regions, while simultaneously inferring a state that aligns with the observed signal.

Learning, inference, memory, and emergence in biological and artificial systems are often described using disparate theoretical frameworks ranging from neural field models to recurrent and attention-based architectures. Here we develop a cognitive field theory in which cognition emerges as a collective nonequilibrium phenomenon governed by the infrared organization of adaptive dynamical time scales. Starting from a stochastic cognitive-field equation with homeostatic stabilization and adaptive manifold geometry, we show that collective cognitive dynamics is organized by slowly relaxing infrared modes embedded within a high-dimensional cognitive manifold. Integrating out latent slow-memory sectors generates retarded self-energy feedback and nonlocal memory kernels governing long-time contextual persistence and collective cognitive coherence. We introduce the time-scale density of states (TDOS) as a fundamental descriptor characterizing the distribution of collective relaxation modes underlying inference, memory, and adaptive reasoning. Learning and adaptation continuously reorganize the infrared TDOS, selectively stabilizing weakly damped collective sectors that support contextual organization and recursive collective dynamics. Near criticality, the infrared TDOS generically develops a broad and nearly flat structure associated with the accumulation of slowly relaxing collective modes, producing scale-free temporal organization and enhanced collective coherence. Within this framework, memory formation, adaptive reasoning, and emergent intelligence arise as hierarchical stages of collective infrared dynamical organization.

Triple-negative breast cancer (TNBC) is an aggressive disease with high mortality and limited treatment options, due to its lack of receptors that have targeted therapies available. The tumor microenvironment (TME) plays a critical role in TNBC progression and therapeutic resistance. In this work, we developed a novel mathematical model to describe key cellular interactions within the TNBC TME, informed by current literature and expert input. Our model consists of a system of ordinary differential equations representing five interacting cell populations: M2 macrophages, cancer-associated fibroblasts, TNBC tumor cells, cytotoxic T lymphocytes, and regulatory T cells. We performed global sensitivity analysis to determine which model parameters most strongly influence tumor burden over a clinically-relevant treatment timeframe. The pathways associated with the most-influential parameters correspond to biological mechanisms that are consistent with known and emerging therapeutic strategies in TNBC, including stromal-mediated tumor support. These results highlight key regulatory interactions within the TNBC TME and provide a quantitative framework for hypothesis generation and future investigation of combination treatment strategies.

Automatic extraction of retinal vascular biomarkers from color fundus images (CFI) is crucial for large-scale studies of the retinal vasculature. We present VascX, an open-source Python toolbox that extracts biomarkers from CFI artery-vein segmentations. VascX starts from vessel segmentation masks, extracts their skeletons, builds undirected and directed vessel graphs, and resolves vessel segments into longer vessels. A comprehensive set of biomarkers is derived, including vascular density, central retinal equivalents (CREs), and tortuosity. Spatially localized biomarkers may be calculated over grids placed relative to the fovea and optic disc. VascX is released via GitHub and PyPI with comprehensive documentation and examples. Our test-retest reproducibility analysis on repeat imaging of the same eye by different devices shows that most VascX biomarkers have moderate to excellent agreement (ICC > 0.5), with important differences in the level of robustness of different biomarkers. Our analyses of biomarker sensitivity to image perturbations and heuristic parameter values support these differences and further characterize VascX biomarkers. Ultimately, VascX provides an explainable and easily modifiable feature-extraction toolbox that complements segmentation to produce reliable retinal vascular biomarkers. Our graph-based biomarker computation stages support reproducible, region-aware measurements suited for large-scale clinical and epidemiological research. By enabling easy extraction of existing biomarkers and rapid experimentation with new ones, VascX supports oculomics research. Its robustness and computational efficiency facilitate scalable deployment in large databases, while open-source distribution lowers barriers to adoption for ophthalmic researchers and clinicians.

Recent work suggests that large-scale, multi-animal modeling can significantly improve neural recording analysis. However, for functional calcium traces, existing approaches remain task-specific, limiting transfer across common neuroscience objectives. To address this challenge, we propose \textbf{CalM}, a self-supervised neural foundation model trained solely on neuronal calcium traces and adaptable to multiple downstream tasks, including forecasting and decoding. Our key contribution is a pretraining framework, composed of a high-performance tokenizer mapping single-neuron traces into a shared discrete vocabulary, and a dual-axis autoregressive transformer modeling dependencies along both the neural and the temporal axis. We evaluate CalM on a large-scale, multi-animal, multi-session dataset. On the neural population dynamics forecasting task, CalM achieves competitive performance against strong specialized baselines after pretraining. With a task-specific head, CalM further adapts to the behavior decoding task and achieves superior results compared with supervised decoding models. Moreover, linear analyses of CalM representations reveal interpretable functional structures beyond predictive accuracy. Taken together, we propose a novel and effective self-supervised pretraining paradigm for foundation models based on calcium traces, paving the way for scalable pretraining and broad applications in functional neural analysis. Code is released at this https URL.

Modern optical microscopes are fully motorised; however, transforming them into truly smart systems requires real-time adjustment of acquisition settings in response to detected objects and dynamic biological events. At the core are classification algorithms that commonly depend on customised software and are generally designed for narrowly-defined biological applications. In addition, they often require substantial annotated datasets for effective training. We introduce a semi-supervised generative adversarial network (SGAN) for robust cell-cycle stage classification under low-resource conditions, adaptable to diverse cellular structures. The framework combines unlabelled microscopy images with synthetically generated samples to mitigate limited annotation, while preserving stable performance even when the unlabelled subset is class-imbalanced. Tested on the Mitocheck dataset, which features five mitosis classes, the model achieved $93 \pm 2\%$ accuracy using only 80 labelled per class and 600 unlabelled images. The proposed algorithm is generic and can be readily adapted to new labeling schemes, classification targets, cell lines, or microscopy modalities through transfer learning. SGAN is well suited for integration into automated microscopes, enabling efficient and adaptable image analysis across diverse biological and microscopy applications.

How do we measure genuine understanding in artificial cognitive systems? Current approaches face a measurement gap: probabilistic systems refine confidence gradually, practice-based systems compile knowledge through repeated execution, and neural systems distribute understanding across opaque embedding spaces. We propose that making understanding measurable requires architectures where understanding formation produces discrete, inspectable structural signatures. This paper presents hierarchical automata built from finite state machines representing patterns and higher-order automata representing compositions. Constrained inference constructs automata from single observations. Similarity detection clusters related automata, making concept robustness quantifiable. Graph memory makes compositional knowledge directly inspectable. Metacognitive mechanisms enable observable reconfiguration. We demonstrate understanding measurement in a simple geometric domain. Graph evolution tracking reveals five measurable signatures: immediate representation formation, structural knowledge, generalization capacity, compositional awareness, and metacognitive access. These measurements distinguish structural understanding from statistical correlation. Our contribution is a framework for making understanding measurable through discrete compositional knowledge structures. This measurement capability complements perceptual learning in neural systems and task execution in neurosymbolic architectures.

Landscape dynamics are key drivers of the movement and distribution of sylvatic hematophagous disease vectors and their (wild) animal hosts. Their habitats are undergoing increasing change, particularly fragmentation, through anthropogenic activity. In this article, we present and analyse a novel mathematical model that explicitly combines anthropization-induced landscape dynamics with the population dynamics of hematophagous vectors and (wild) animals dynamics. We develop a phenomenological and analytically tractable two-patch model in which the migration terms between the patches nonlinearly depend on the anthropization level of the patches. Our model analysis comprising analytical stability analysis and numerical bifurcation analysis provides information on how changes in model parameters, especially anthropization levels, shape the long-term dynamics in the model. Precisely, we find that low anthropogenic activity allows for a vector-animal coexistence state, while high anthropization leads to a vector extinction state. However, we establish that for intermediate anthropization levels, the transition between the two states is not necessarily monotonic, but may instead occur via a sequence of concurrent bifurcations along the anthropization axis.

The logical chain of this paper proceeds as follows: differential stability leads to the spontaneous emergence of information, which enables the physical selection of RNA, followed by compartmentalization as a computational platform, then non-genetic information accumulation in metabolic networks, ribosomal assembly from cross-catalytic modules, and ultimately the co-origin and coexistence of cells and viruses. Each link in this chain constitutes the premise for the next, and each transition is driven by the same underlying principle, namely, selective enrichment via stability differences, operating under progressively more complex boundary conditions. The aim of this paper is to demonstrate that, under plausible early Earth physicochemical conditions, the entire transition from random chemistry to genetic systems can be derived through a unified, logically necessary mechanism, without recourse to any ultra-low-probability chance events. If this argument holds, then the origin of life is no longer an inscrutable "fortuitous miracle" but the "inevitable emergence" of a complex chemical system under specific boundary conditions.

Spatial transcriptomics (ST) provides essential spatial context by mapping gene expression within tissue, enabling detailed study of cellular heterogeneity and tissue organization. However, aligning ST data with histology images poses challenges due to inherent spatial distortions and modality-specific variations. Existing methods largely rely on direct alignment, which often fails to capture complex cross-modal relationships. To address these limitations, we propose a novel framework that aligns gene and image features using a ranking-based alignment loss, preserving relative similarity across modalities and enabling robust multi-scale alignment. To further enhance the alignment's stability, we employ self-supervised knowledge distillation with a teacher-student network architecture, effectively mitigating disruptions from high dimensionality, sparsity, and noise in gene expression data. Extensive experiments on seven public datasets that encompass gene expression prediction, slide-level classification, and survival analysis demonstrate the efficacy of our method, showing improved alignment and predictive performance over existing methods.

Wearable movement data is collected by nearly all commercially available smartwatches and is a valuable resource for mental health research, reflecting fine-grained temporal behavioral trends. Despite its promise, the development of foundation models for health wearable modeling remains limited when compared to clinical image and text analysis. We designed transformers with patch embeddings and used self-supervised masked autoencoder pretraining on minute-level week-long actigraphy (physical activity intensity measurement) sequences to develop and evaluate the Pretrained Actigraphy Transformer (PAT). PAT is an open-source foundation model for wearable movement time series that combines week-long temporal modeling, psychiatric outcome evaluation, and reproducibility on public data. Pretrained on data from 21,538 U.S. participants in a nationally representative cohort from the National Health and Nutrition Examination Survey (NHANES), PAT consistently outperformed non-foundation-model baselines across mental health prediction tasks-including benzodiazepine and SSRI use, depression, and sleep abnormalities. During the benzodiazepine medication usage prediction task, PAT demonstrated the largest improvement over non-foundational deep learning models commonly used for time-series modeling (i.e., 55.6% improvement over the LSTM, 21.4% improvement over the 1-D CNN, 14.8% improvement over the ConvLSTM). Beyond predictive accuracy, PAT provides interpretable attention maps highlighting specific periods of daily activity most important for clinical predictions, offering model transparency and potential clinical insights. The results suggest that PAT offers an easy-to-deploy, adaptable and scalable solution to advance clinical insight from wearable sensor data for researchers and clinicians. GitHub: this https URL

The integration of heterogeneous multi-omics datasets at a systems level remains a central challenge for developing analytical and computational models in precision cancer diagnostics. This paper introduces Multi-Omics Graph Kolmogorov-Arnold Network (MOGKAN), a deep learning framework that utilizes messenger-RNA, micro-RNA sequences, and DNA methylation samples together with Protein-Protein Interaction (PPI) networks for cancer classification across 31 different cancer types. The proposed approach combines differential gene expression with DESeq2, Linear Models for Microarray (LIMMA), and Least Absolute Shrinkage and Selection Operator (LASSO) regression to reduce multi-omics data dimensionality while preserving relevant biological features. The model architecture is based on the Kolmogorov-Arnold theorem principle and uses trainable univariate functions to enhance interpretability and feature analysis. MOGKAN achieves classification accuracy of 96.28 percent and exhibits low experimental variability in comparison to related deep learning-based models. The biomarkers identified by MOGKAN were validated as cancer-related markers through Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis. By integrating multi-omics data with graph-based deep learning, our proposed approach demonstrates robust predictive performance and interpretability with potential to enhance the translation of complex multi-omics data into clinically actionable cancer diagnostics.

Recent developments in the Internet of Bio-Nano-Things (IoBNT) are laying the foundation for innovative healthcare applications that envision a network of remotely coordinated nanodevices within the human body to monitor and actuate over potential diseases. However, interconnecting such nanodevices requires communication strategies that can cope with molecular communication (MC) channels, whose complex, stochastic, and dynamic behavior often makes accurate physical modeling infeasible. To explore the limits of nanodevice interconnectivity under these conditions, this survey focuses on data-driven communication strategies for MC systems, with particular emphasis on machine learning (ML) methods and neural network (NN) architectures for a robust and adaptive communication scheme at the nanoscale. Research on NN-enabled MC spans several aspects covered in this survey, including NNs for communication in IoBNT networks, the feasibility of biocompatible NN realization, explainable approaches, and the generation of training datasets. We also include open-source code examples to support reproducible research across key MC scenarios. Finally, we identify emerging challenges, including the need for robust NN architectures, biologically integrated NN modules, and scalable training strategies.

We investigate a relatively under-explored class of hybrid neurosymbolic models that integrate symbolic learning with neural reasoning to construct data generators meeting formal correctness criteria. In Symbolic Neural Generators (SNGs), symbolic learners examine logical specifications of feasible data from a small set of instances -- sometimes just one. Each specification in turn constrains the conditional information supplied to a neural-based generator, which rejects any instance violating the symbolic specification. Like other neurosymbolic approaches, SNG exploits the complementary strengths of symbolic and neural methods. The outcome of an SNG is a pair $(H, X)$, where $H$ is a symbolic description of feasible instances constructed from data, and $X$ a set of generated new instances that satisfy the description. We introduce a semantics for such systems, based on the construction of appropriate base and fibre partially-ordered sets combined into an overall partial order. We implement an SNG combining a restricted form of Inductive Logic Programming (ILP) with a large language model (LLM) and evaluate it on early-stage drug design. Our main interest is the description and the set of potential inhibitor molecules generated by the SNG. On benchmark problems -- where drug targets are well understood -- SNG performance is statistically comparable to state-of-the-art methods. On exploratory problems with poorly understood targets, generated molecules exhibit binding affinities on par with leading clinical candidates. Experts further find the symbolic specifications useful as preliminary filters, with several generated molecules identified as viable for synthesis and wet-lab testing.

Human vision is foveated, with variable resolution peaking at the center of a large field of view; this reflects an efficient trade-off for active sensing, allowing eye-movements to bring different parts of the world into focus with other parts of the world in context. In contrast, most computer vision systems encode the visual world at a uniform resolution, raising challenges for processing full-field high-resolution images efficiently. We propose a foveated vision interface (FOVI) based on the human retina and primary visual cortex (V1), that reformats a variable-resolution retina-like sensor array into a uniformly dense, V1-like sensor manifold. Receptive fields are defined as k-nearest-neighborhoods (kNNs) on the sensor manifold, enabling kNN-convolution via a novel kernel mapping technique. We demonstrate two use cases: (1) an end-to-end kNN-convolutional architecture, and (2) a foveated adaptation of the DINOv3 ViT foundation model, leveraging low-rank adaptation (LoRA). These models provide competitive performance with a fraction of the pixels and computational cost of full resolution non-foveated baselines, opening pathways for efficient and scalable active sensing for high-resolution egocentric vision. Code (this https URL) and pre-trained models (this https URL) are available.

Protein sequences are abundant in repeating segments, both as exact copies and as approximate segments with mutations. These repeats are important for protein structure and function, motivating decades of algorithmic work on repeat identification. Recent work has shown that protein language models (PLMs) identify repeats, by examining their behavior in masked-token prediction. To elucidate their internal mechanisms, we investigate how PLMs detect both exact and approximate repeats. We find that the mechanism for approximate repeats functionally subsumes that of exact repeats. We then characterize this mechanism, revealing two main stages: PLMs first build feature representations using both general positional attention heads and biologically specialized components, such as neurons that encode amino-acid similarity. Then, induction heads attend to aligned tokens across repeated segments, promoting the correct answer. Our results reveal how PLMs solve this biological task by combining language-based pattern matching with specialized biological knowledge, thereby establishing a basis for studying more complex evolutionary processes in PLMs.

Decoding natural language from non-invasive EEG signals is a promising yet challenging task. However, current state-of-the-art models remain constrained by three fundamental issues: Semantic Bias, where outputs collapse into generic linguistic templates; Signal Neglect, where models rely heavily on LLM priors to hallucinate fluent text even in the absence of meaningful signals; and the "BLEU Trap", where high-frequency stopwords inflate n-gram metrics, masking a lack of true semantic fidelity. To resolve these challenges, we move beyond conventional end-to-end pipelines and propose SemKey, a novel multi-stage framework that enforces signal-grounded generation through four decoupled semantic objectives: sentiment, topic, length, and surprisal. We extract these semantic anchors from EEG embeddings directly, then unify them with an Active Retrieval Decoding mechanism, compelling the LLM to ground its token generation in the neural signals rather than defaulting to linguistic priors. Furthermore, we break the BLEU Trap by establishing a comprehensive evaluation protocol using rigorous retrieval and distribution-based metrics such as Fréchet Distance. Extensive experiments demonstrate that SemKey effectively mitigates hallucinations on noise inputs and achieves SOTA performance on these robust protocols. Code will be released upon acceptance at this https URL.