Papers

Linking individual-level processes, such as survival and reproduction, to ecosystem dynamics is challenging due to interactions among populations and species. These interactions depend on organism traits, such as age, body size, and mass, and are influenced by fluctuations in population structure. Integral projection models (IPMs) have advanced our understanding of dynamics arising from population structure in continuous traits. Yet, these models have only been applied to relatively simple systems, featuring one or two species and spanning a fraction of the trophic levels observed in natural food webs. We extend these models in four ways. First, we provide a general multi-species IPM where population structure can be included in primary producers and primary and secondary consumers. Second, we link individual-level processes to ecosystem functioning by mapping fluxes of biomass between species to fluxes within individuals, using bioenergetic principles for allocation to maintenance, reproduction, or growth. Third, we introduce a nutrient recycling loop through which decomposers break down organic matter into nutrients that enable plant growth. Finally, we provide an efficient fitting algorithm to calibrate the model with time series of population sizes. We showcase our approach by parameterising and fitting the model to counts of elk, bison, wolves, and cougars in northern Yellowstone National Park to study the impact of predator extirpation and recovery. Our model predicts that bison decrease following predator removal and recover following predator recovery, suggesting a positive contribution of wolves and cougars to the increase of bison in northern Yellowstone. It also reproduces indirect effects of predators on woody deciduous vegetation regeneration. Our approach is suitable for a broad range of ecosystems, with applications for population management and ecological forecasting.

Successful discovery, integration and reuse of data relies on the availability of rich, well-structured and machine-readable metadata to describe every aspect of the data, from sample sources to collection processes to experimental protocols. The use of standardised terminologies to express concepts in a harmonised fashion lies at the core of high-quality data annotation, increasing the FAIRness of the data, facilitating data integration and promoting reproducibility. Here, we describe the Human Ancestry Ontology (HANCESTRO), originally developed to improve standardised reporting of genetic ancestry genomic resources such as the NHGRI-EBI GWAS Catalog and the Human Cell Atlas through high-level population descriptors, and more recently expanded to include diverse and previously under-represented populations in genomics and genetics research. HANCESTRO provides a framework for population descriptors that includes both ancestry based on the analysis of genetic information and self-reported ethnicity, which is based on social and cultural factors that don't necessarily align with genetic populations. By enabling the accurate and interoperable representation of population-related data, it promotes inclusive, representative and reproducible science.

Polymer-functionalised mesopores are an emerging technology for colloid separation, sensing and delivery. Their potential is strikingly illustrated in living cells, where nuclear pore complexes (NPCs) control biocolloid transport between the nucleus and the cytosol. Even colloids much smaller than the biopolymer-filled NPC channel are effectively blocked, but some larger colloids with distinct surface features rapidly permeate. Simplistically, one may expect any polymer filling to obstruct and slow down colloid transport. We demonstrate how a polymer filling that attracts colloids and extends beyond the mesopore, thus maximizing colloid capture, can instead increase permeation compared to a bare pore. We also define how polymer-filled mesopores can effectively gate colloids according to their size and surface features. Our findings provide a basic physical explanation for the exquisite permselectivity of NPCs, and a rational design strategy for novel mesopore-based separation, sensing, catalysis and drug delivery devices with enhanced performance features.

Linking the internal representations of deep neural networks (DNNs) to human mental representations is important for using DNNs as computational models of human vision. Existing DNN representations remain insufficiently similar to human mental representations, which are not directly observable and are therefore commonly measured through large-scale similarity judgments of object images. A natural approach to narrowing this gap is to directly transfer the relational structure of human representations into DNNs, and previous studies have reported improved human-DNN representational similarity. However, whether this improvement holds under stricter evaluation remains untested in two respects: fine-grained alignment at the individual-object level, and generalization to a human embedding derived from a dataset independent of the training data. Here, we employ an unsupervised comparison method, Gromov-Wasserstein optimal transport (GWOT), which estimates human-DNN correspondences from the internal distance structure alone and thereby tests fine-grained alignment. We further assess generalization on a curated test set of concepts non-overlapping with the training data. We show that fine-tuning pre-trained DNNs with Relational Knowledge Distillation (RKD), an established relational transfer method, brings DNNs close enough to humans to be aligned at the individual-object level on this test set. We also show that this improvement is driven by a more human-like global structure, as reflected in the ordering of distances among coarse categories, while the local human-DNN nearest-neighbor overlap rate remains largely unchanged. These findings indicate that relational transfer from humans brings the global structure of pre-trained DNNs close enough to the human structure to enable fine-grained human-DNN alignment without supervision.

Biased migration up chemical gradients and proliferation are fundamental drivers of collective invasion in several biological processes ranging from embryonic morphogenesis to cancer. Nonetheless, our understanding of how their interplay yields distinct invasion patterns remains incomplete. In this work, we propose a multiscale framework to systematically derive macroscopic hydrodynamic theories of cell invasion from a mesoscopic description of cells as biased self-propelled, interacting particles that proliferate. Our framework reveals how clump and stream invasion patterns emerge from the same kinetic equation under different asymptotic regimes of cell proliferation. Stream invasion is characteristic of cell populations in which proliferation balances cell motion. In contrast, clump invasion requires a separation of the hydrodynamic timescale of motion and the slower timescale of proliferation. By means of a multiple-scale approach, our analysis reveals that clump invasion is described as a slow evolution through a family of mass-dependent travelling-wave solutions. Overall, our work offers a novel approach to investigate multiscale regulation of cell invasion in systems where cell proliferation and collective invasion evolve on distinct timescales.

Banana crop diseases threaten food security across the world, yet field diagnosis remains difficult because of limited expert access and visual similarity among disease classes. We present BananaVLM, a domain-adapted vision-language model for banana disease diagnosis built on LLaVA-v1.5-7B and fine-tuned with Low-Rank Adaptation (LoRA). To address the lack of multimodal agricultural instruction data, we introduce BananaInstruct, an automated pipeline that converts raw disease images into $\approx$80,000 question–answer pairs without manual annotation, using LLaVA-1.5-13B for symptom descriptions and Mistral-7B for agricultural Q&A generation and label grounding. Evaluated against 14 open-source and 5 closed-source VLMs on a nine-class benchmark, BananaVLM achieves 92.21% in-domain and 83.28% out-of-domain (OOD) classification accuracy, outperforming the best closed-source baseline, Gemini~2.5~Pro, by 49.8 and 63.3 percentage points, respectively. Binary healthy/diseased identification reaches 98.38% OOD accuracy with 1.00 recall. A controlled LoRA–DoRA comparison shows that DoRA performance degrades across epochs, while LoRA remains more effective for fine-grained classification. Qualitative evaluation using nine LLM judges yields G-Eval win rates of 0.81–0.96, while five domain experts prefer BananaVLM in 98.48% of 990 blind pairwise comparisons. These results show that lightweight domain adaptation with automated instruction tuning is an effective and scalable approach for specialized agricultural AI. Code, datasets, and model weights are available at https://github.com/samy101/banana-vlm

Orchestra composes two independently built bioinformatics MCP servers – RegNetAgents, which infers gene regulatory network topology from ARACNe networks, and CASCADE, which supplies four independent evidence sources (LINCS knockdown, DepMap essentiality, super-enhancer status, DoRothEA transcription-factor confidence) – into one multi-agent workflow exposed via the Model Context Protocol. Its central architectural claim is that requiring RegNetAgents' topology evidence and CASCADE's experimental evidence to agree on a candidate regulator yields a more trustworthy candidate than either alone – not previously tested directly, since RegNetAgents' own validation asked only whether its candidate lists beat chance. We test this on the TCGA tumor-acquired regulator tier (regulators in a gene's tumor ARACNe network but absent from the GREmLN population-averaged baseline), selecting candidates by ARACNe mutual-information (MI) edge weight. On RegNetAgents' published BRCA/COAD focal-gene panel plus matched negative controls, agreement among at least 2 of the 4 CASCADE sources predicts OncoKB cancer-gene status among focal genes (odds ratio 2.89, Benjamini-Hochberg-adjusted p=0.0166) but not among negative controls (p=0.0721); a single source is not diagnostic for either group. The pattern replicates and strengthens in a third cancer type, STAD, on a separately constructed panel (odds ratio 5.82), and against an independently curated ground truth (the Sanger COSMIC Cancer Gene Census). MI edge weight is the strongest single predictor overall (p=0.0003); a logistic-regression likelihood-ratio test confirms corroboration adds value beyond it in both panels (p=0.0234; p=0.0001). Every experiment invokes Orchestra's real agentic entry point.

We study long-run optimal intervention strategies for endemic infections in a stochastic susceptible-infected-susceptible (SIS) model. The proportion of infected individuals evolves as a diffusion process with random fluctuations, while a control variable $\zeta_t\in[0,\tilde{\zeta}_{max}]$ represents the intensity of public health interventions that reduce transmission for some intervention threshold $\tilde{\zeta}_{max}\in (0,1]$. The objective is to minimize the long-run average societal cost, balancing the burden of infection against the costs of interventions. Under a concave intervention cost structure the problem can be formulated as an ergodic stochastic control problem, whose structure implies (under certain additional conditions) that optimal interventions are of bang-bang type, switching between no intervention and the maximal admissible intervention at a single switching threshold in the infection level. We construct candidate value functions, rigorously verify optimality in this single-threshold case, and relate the results to extinction and persistence properties of the underlying SIS dynamics in the absence of control. In our framework, the analysis provides a rigorous justification for the threshold-based intervention rules commonly used in epidemic management.

Sparse autoencoders (SAEs), and other interpretability methods could turn AI models in Biology and other fields into engines of scientific discovery by explaining the superhuman capabilities of those models. However, a latent is only useful if we know three things: whether it is coherent, whether it can be described, and whether that description has predictive power. These questions are routinely conflated. We assemble them into a single pipeline and report the practical innovations each stage required. First, cross-seed dictionary stability prioritises which latents are worth spending resources to investigate. Second, an intruder-detection task asks whether a latents activating examples share a recognizable pattern. Third, a separate pass proposes a candidate biological description which we convert into falsifiable predictions which can be tested in silico. Deployed on the Boltz-1 Pairformer trunk, stability prioritisation finds interpretable latents using about 4.4 times fewer latent evaluations each, and at 5.2 times lower measured cost, while recovering over half of them, and the external check shows the surfaced motifs are significantly enriched for their claimed annotations. The results also suggest a possible tension: the cross- seed stability might be selecting for some types of features, like structure-related ones, much more than others, such as function-related features.

Biological heterogeneity in Alzheimer's Disease (AD) poses a critical diagnostic challenge, particularly for traditional linear methods that fail to capture non-linear neural dynamics. To address this, we propose a diagnostic framework utilizing the Large Brain Model (LaBraM), pretrained on over 2,500 hours of EEG data. By integrating these high-dimensional latent embeddings with a non-linear Random Forest classifier, our approach effectively isolates robust disease markers. Under a rigorous subject-independent 5-fold cross-validation protocol, the method achieves an ROC-AUC of 89.36% +/- 3.49%, PR AUC of 81.45% +/- 4.43%, and Balanced Accuracy of 82.44% +/- 4.34% in distinguishing dementia patients from healthy controls. Notably, this performance uses only 8-second EEG segments, surpassing traditional spectral baselines, including band-power and parameterized oscillatory features (FOOOF). Post-hoc occlusion analysis confirms the model captures clinically validated biomarkers, specifically occipital-frontal Alpha and Theta rhythm degradation. Additional neurophysiological alignment analysis demonstrated that higher LaBraM-predicted dementia probability significantly correlated with worse cognitive performance, greater clinical severity, increased theta and alpha relative power, and higher aperiodic exponent. These findings demonstrate that deep latent representations extract clinically relevant signatures from noisy signals, enabling precise, rapid, and data-efficient diagnosis.

The evolutionary fate of proteins is driven by both folding stability and biological function, dual constraints that often conflict, creating frustration and imposing functional costs beyond stability. These costs can be captured by a "dark energy": the difference between the evolutionary energy of protein sequences and their physical folding energy. Recent advances in deep mutational scanning, protein language models, and inverse-folding models have enabled the quantification of dark energy across the protein universe. We review the computational and experimental approaches that disentangle folding and function at scale, revealing a dark energy component and providing new insights into how biological information flows from sequence to structure to function and back to sequence.

Recent research grounded in experiments has connected neural ring models of vision to how animals navigate a complex landscape of attractive targets. In this paper we investigate the mathematical and biological implications of a three-stage model where animals pre-process visual stimuli to identify a discrete set of targets, process this input to select the dominant targets, and then post-process this information to navigate the landscape. Incorporating finite target sizes and a neural density allows consistent target acquisition and pursuit reflecting what is seen in nature. Mathematically, we show this model corresponds to an energy minimization problem, simplifying both its analysis and numerical implementation. Biologically, we argue that the model is physiologically motivated, reproduces experimental observations, and presents a fairly direct pathway to decoding neural geometry via empirical data. Our results also demonstrate an analysis pipeline for future model extensions and suggest specific avenues for experimental research to test and parameterize our quantitative framework.

Ancient DNA and new inference methods have transformed the study of human origins, but consensus has not followed. Evidence increasingly indicates that hominin populations were pervasively structured and admixed, so complexity rather than simplicity is the appropriate prior. Here I highlight two blind spots that impede resolving that complexity. First, every inference passes through summaries of the data, and those summaries bound what can be recovered. Second, the space of candidate models is vast, yet competing model classes are rarely fit to common data, so a reported best model carries little evidence about untested model classes. This second blind spot reflects practice rather than data. It can be narrowed by testing competing models against withheld summaries and by reporting the models that were tried and rejected rather than only the winner.

Motivation: Accurate prediction of mutation-induced protein–protein binding free-energy changes is important for antibody affinity maturation, yet scarce labels and complex interface geometry limit generalization. Heterogeneous predictors may process three-dimensional complexes without preserving the cross-chain signals most relevant to a mutation in their final scalar output. Results: We introduce CIR-DDG, a lightweight residual adapter that combines a fixed base prediction with 22 interpretable descriptors of cross-chain distance, contact density and site–partner context. In complex-level five-fold evaluation on SKEMPI 2.0 measurements from 343 complexes, CIR-DDG improved all six tested backbones on antibody–antigen interface mutations: Spearman correlation increased by 0.0346–0.1296, while RMSE decreased by 0.0074–0.0408\kcalmol. Cross-validated probing, equal-capacity controls and feature ablations support the complementarity of explicit geometry. On an independent SARS-CoV-2 RBD–ACE2 deep-mutational-scan benchmark of 3669 substitutions, the fold-specific adapters transferred without any retraining: the absolute interface Spearman correlation increased by 0.026–0.081 for all four evaluable backbones, showing that the learned geometric correction generalizes beyond SKEMPI thermodynamic measurements. Availability and implementation: CIR-DDG is available at https://github.com/ecnuabmlab/CIR-ddG.

Large-scale, standardized datasets have driven many advances in AI-based scientific modeling, from protein structure prediction to natural language processing. Infectious disease epidemiology is increasingly adopting AI methods for forecasting, surveillance, and outbreak analytics, but the time-series data available to train them remains orders of magnitude smaller than the corpora behind the advances seen in other fields. Because the scope of real-world epidemiological data cannot practically reach the scale needed to train truly large-scale AI methods, simulated data provides a possible alternative. Here we introduce Reservoir, a large open simulator and dataset of realistic epidemic simulations in which every trajectory carries complete ground-truth labels, including quantities that cannot be measured directly in a real outbreak, such as true infection counts, time-varying reproduction numbers, and counterfactual intervention effects. Reservoir is generated by a stochastic simulator with realistic noise and reporting artifacts, together with interventions with configurable timing, compliance, and age-dependent efficacy. The current release contains 500,000 outbreak trajectories spanning one billion simulated days across diverse pathogen characteristics, population structures, and intervention regimes. Reservoir enables counterfactual experiments, surveillance-design studies, and training of epidemic models at a scale real-world datasets cannot provide.

Ordinary differential equations are commonly used throughout the sciences to build mechanistic models of time-dependent processes. Often, such models are Markov models describing the time-evolution of different interconnected "states". When these models have no "sources" or "sinks", they naturally conserve the total population of the system. When each state is reachable (directly or indirectly) from any other state, these models are called irreducible. For many applications, a deterministic system of ordinary differential equations (ODE) is the most suitable modelling approach. We summarise important mathematical results which show that this irreducibility property guarantees the existence and uniqueness of global stable equilibria, and discuss the computationally-efficient implementation of such ODE-based models. We also discuss the condition of microscopic reversibility and show how it guarantees non-oscillatory behaviour, which enables additional efficiencies in computations. These properties and methods are demonstrated through example models of biological phenomena, where we demonstrate their importance for efficient model fitting and simulation.

Biological reaction networks often exhibit complex transient behavior that cannot be explained solely by the analysis of steady states or long-term persistence. Existing structural approaches identify persistent system properties and have considered transitions between organizations, but do not provide a general criterion for admissible transitions between arbitrary species configurations. We introduce admissible transitions, a new structural concept that describes feasible changes between species subsets using only reaction network structure and feasible reaction fluxes, independently of kinetic parameters. We prove that solutions of reaction-based ordinary differential equation systems induce canonical sequences of admissible transitions with a fundamental asymmetry: closure-building transitions are uniquely determined by network structure, whereas downward transitions are generally non-unique and depend on the realized trajectory. This establishes a structural layer linking network topology to transient system evolution. The framework is illustrated using a classical HIV immune-response model. By extending structural reaction network analysis from persistent states to transient dynamics, the proposed theory provides a general, kinetics-independent framework for analyzing reachability, organization formation, and transient behavior in biological systems.

This paper establishes that the system of ordinary differential equations arising from the sequential and distributive dual futile cycle has the structural capacity for Hopf bifurcations, whenever it is endowed with general parameter-rich kinetics, but it loses such capacity if it is endowed with mass action kinetics. The proof of the latter fact relies on a Routh-Hurwitz approach, improved by few preliminary structural considerations, but a decisive contribution came from ChatGPT Sol 5.6, which provided a nontrivial positivity certificate. A second purpose of the paper is indeed to document such AI-contribution, exploiting this well-studied toy-model, and to offer an example of the utilizability of such tools in the context of computer algebra and positivity certificates for large polynomials.

Many volume-based neuroimaging workflows require NIfTI label volumes, whereas a growing number of cortical parcellations are distributed in CIFTI surface space. We present Surf_2_Volume, a Python package for converting categorical CIFTI cortical parcellations into NIfTI volumes by resampling surface labels and filling a target cortical ribbon. The package provides configurable control over cortical-ribbon support and allows users either to specify the mapping threshold manually or to select suitable thresholds automatically according to the similarity between parcel-size distributions in the converted volume and the source CIFTI, without requiring a volumetric reference for threshold selection. We evaluated Surf_2_Volume using the Schaefer2018 100-parcel 7-network and 17-network atlases and compared its automatically selected outputs with multiple Connectome Workbench surface-to-volume mappings. Surf_2_Volume produced more continuous cortical coverage while avoiding the substantial underfilling observed with conservative Workbench mappings and the wider spatial expansion observed with more permissive nearest-vertex mappings. Across both atlases, the automatically selected Surf_2_Volume outputs showed consistently closer agreement with the paired reference volumes than the tested Workbench mappings. In HCP resting-state data, parcel time series extracted from Surf_2_Volume volumes also showed consistently high agreement with those obtained from the paired reference volumes. Together, these results support Surf_2_Volume as a configurable Python-based solution for converting CIFTI cortical parcellations to NIfTI volumes, with automatic threshold selection and stable performance across the two tested network parcellations.

Reproducible biomarkers that monitor motor and cognitive dysfunction in amyotrophic lateral sclerosis (ALS) are needed. Previous work reported reduced beta-band event-related desynchronization (ERD) and attenuated post-movement event-related synchronization (ERS) in frontal and parietal channels during the sustained attention to response task (SART). We aimed to validate these findings in Dutch and Irish cohorts. A randomized SART with 128-channel electroencephalography (EEG) was performed in Dutch (ALS n=63; matched controls n=64) and Irish (ALS n=36; controls n=36) cohorts. Non-phase-locked activity was quantified using complex Morlet wavelets to compute baseline-normalized post-stimulus ERD and ERS. Group differences and EEG associations with task performance were examined for Go and NoGo trials. In both control cohorts, the SART elicited theta ERS and alpha- and beta-band ERD, followed by rebound beta ERS across the frontoparietal axis, with greater theta ERS, stronger alpha ERD, and weaker beta ERS on NoGo than Go trials. In contrast to previous null findings, people with ALS showed reduced Go and total accuracy in both centres (p<0.002) and slower Go responses in the Irish cohort (p=0.02). The ALS cohorts differed in NoGo accuracy, Go response time, and anticipation errors (all p=0.03). Replicating prior work, the Irish ALS cohort showed reduced beta ERD and ERS versus controls, with no such differences in the Dutch cohort. Greater beta ERS was associated with faster Go responses in both centres (p<0.011). ERD and ERS changes replicated in the Irish but not the Dutch cohort. Cohort differences in these metrics reflected, at least in part, varying functional impairment. Reproducible control findings and consistent associations with motor performance indicate reliable capture of cortical motor network dysfunction, supporting these measures as EEG markers of motor decline in ALS.

Objective: Choosing between noninvasive respiratory support (NIRS) and invasive mechanical ventilation (IMV) for acute respiratory failure is a time-sensitive decision. Although guidelines provide population-level guidance, it remains unclear who benefits more from NIRS than IMV. The standard outcome, ventilator-free days at 28 days (VFD-28), scores death and prolonged ventilation equally, and current estimators do not distinguish between them. We developed and validated a censoring-aware Digital Twin framework for Individualized Treatment Effects of Non-Invasive Respiratory Support (DINIRS) to estimate individualized treatment effects (ITEs) that capture both mortality and ventilation duration. Materials and Methods: We emulated a target trial in 5,336 MIMIC-IV patients and trained DINIRS on 23 baseline clinical variables measured during the first 24 ICU hours. A transformer encoder with a survival attention gate decomposed VFD-28 into survival probability and conditional ventilation duration. A cross-fitted, doubly robust learner estimated ITEs. We externally validated DINIRS in 2,540 patients from the multi-site eICU-CRD dataset without retraining. Results: The DINIRS policy achieved a mean benefit of 2.07 ventilator-free days per patient (207 per 100 patients) compared with observed practice. Predicted NIRS benefit was higher among patients with less organ dysfunction (88.4% versus 49.0%) and persisted across hypoxemia severity. External validation reproduced this pattern. Discussion: The NIRS benefit stemmed from shorter ventilation among survivors rather than from reduced mortality, indicating that avoiding intubation-associated complications was the primary mechanism. Conclusion: Prospective validation is needed before these estimates inform treatment decisions. The decomposition framework can be extended beyond respiratory support to any zero-inflated composite outcome.

Protein language models organize sequence and structure at scale, but a global representation of how proteins respond to mutation remains lacking. We present RegimeFormer, a large protein perturbation model coupled to RegimeAtlas, constructed by harmonizing and indexing 202,556,313 non-redundant protein sequences across the tree of life. A diversity-preserving one-million-protein subset provides the high-resolution training and inference layer, with 995,995 proteins yielding residue-level summaries across 407,048,356 residues and substitution-specific predictions available on demand. Across experimental deep mutational scanning, molecular benchmarks, structural confidence and evolutionary constraint, RegimeFormer identifies reproducible protein-level perturbation regimes that organize residue fragility, adaptability and predictive uncertainty. Regime conditioning improves substitution-specific prediction, with the largest relative gains under unseen-protein, unseen-family and low-homology evaluation. RegimeFormer-derived molecular priors further improve downstream transcriptomic and drug-response modelling. Together, RegimeFormer and RegimeAtlas provide a scalable framework for mapping, predicting and querying protein perturbation landscapes across global sequence space.

Masked discrete diffusion models perform strongly on text, code, and biological sequences, but their training objective rewards only naturalness, and retraining the generator for every new reward is expensive. Inference-time steering of a frozen model either guides the sampler by the reward gradient or searches over several trajectories, and recent samplers combine the two. Such combinations are assembled as pipelines that leave three choices at their defaults: a guidance estimate resting on one Gumbel draw per sample, a reward tilting placed without reference to the distribution the combination then targets, and a selection temperature held fixed although the spread of per-step rewards drifts. We identify that distribution and settle the three choices against it. We therefore propose Variance-reduced Guidance and Adaptive Selection (VGAS), a simple yet effective inference-time framework that reduces the variance of the guidance estimate for both reward types, applies the reward tilting in the clean-token logits, where the pretrained schedule is preserved, and sets the selection temperature per step. Across regulatory DNA, protein and small-molecule benchmarks, VGAS attains the best training-free reward and matches or surpasses a reward-fine-tuned generator.

High-throughput drug screening relies on low-cost primary assays to prioritize compounds for more expensive dose-response profiling, where potency is ultimately quantified. Current screening strategies largely focus on identifying compounds that will confirm biological activity on follow-up, implicitly assuming that confirmed activity will also yield a usable potency estimate. However, confirmed biological activity in screening does not necessarily translate into a quantifiable potency, because active compounds can still fail to produce a reportable dose-response estimate. We therefore present a framework for modeling quantifiability, whether follow-up testing will yield a usable potency estimate, as a distinct triage objective from biological activity. Quantifiability was strongly predictable from the preceding low-cost screen, with most predictive information arising from the observed screening features rather than molecular structure. Response-based predictors remained robust on previously unseen chemical scaffolds and generalized across held-out assay-mechanism families, while the probability of successful quantification varied strongly with response amplitude and assay context. These findings establish experimental measurability, distinct from biological activity, as a predictable property of screening outcomes and show that quantifiability-aware triage can improve the allocation of costly dose-response profiling capacity.

Spreading processes on networks often interact with other dynamics on the same nodes. Neurodegenerative disease provides one example: pathological proteins spread through anatomical connections, while neuronal activity influences and is altered by their spread, forming a spreading-activity feedback loop. However, models coupling pathological protein spreading and neuronal activity have largely focused on linear feedback between the two processes. Here we show that nonlinear feedback can fundamentally change the invasion dynamics in a susceptible–infected–susceptible spreading process coupled to a co-evolving activity process. On general weighted networks, the strength and shape of the feedback may generate finite-amplitude invasion thresholds, hysteresis, and endemic multistability. On regular graphs with homogeneous dynamics, we rule out periodic solutions and show that the degrees of polynomial coupling functions bound the number of endemic states, while monotone couplings require reinforcing feedback for multistability. We test these predictions in simulations of a stochastic spiking neuronal network described by quadratic integrate-and-fire dynamics, where we recover both finite-amplitude invasion thresholds and endemic bistability. These results show that feedback from activity processes can create hysteresis and multistability in network spreading dynamics. In neuroscientific applications, our work suggests that neuronal dynamics may act as a control point in neurodegenerative disease, with even transient changes in activity capable of tipping the brain between health and disease.

Updating a graph by inserting or replacing nodes while preserving semantics and reusing existing structure is a recurring computational problem. In population genetics, this problem arises in the genotype representation graph (GRG), a directed acyclic graph that losslessly encodes phased genetic variation across hundreds of thousands of samples by sharing subgraph structure for individual mutations. In a GRG, each mutation's carrier set is implicitly encoded as the set of leaf nodes reachable from the node it is assigned to. Updating a mutation is therefore a structural editing problem, and current approaches remap mutations individually. This paper introduces a batched mutation-remapping algorithm that replaces independent reuse-aware traversals with a single shared reverse-topological pass, identifying reuse candidates for an entire batch at once. The pass propagates compact bit-parallel per-mutation state and uses an adaptive sparse/dense carrier set representation spanning rare-to-common variant densities. Batching is the memory-scalable complement to split-based parallelism, which instead replicates graph and traversal state per worker. Our remapping is evaluated on a controlled update workload and on end-to-end allele polarization, a bulk carrier set update that is common in population genetic analysis. Our approach is up to 10.5$\times$ faster than independent remapping while preserving exact carrier-set semantics.

In this paper, we study a slow-fast model of Intracellular Calcium Concentrations (ICC) within a stochastic framework, where stochastic perturbations are introduced to capture the intrinsic randomness of such biological systems. We first analyse the three-dimensional single-cell model, based on the FitzHugh-Nagumo structure, and subsequently extend the analysis to a coupled two-cell six-dimensional setting. We prove existence and uniqueness of solutions and establish positivity of the calcium variable. We obtain explicit mean-square stability estimates around deterministic equilibria and quantify how they depend on the noise intensity. Numerical simulations for both the single-cell and coupled systems illustrate relevant qualitative behaviours and reveal additional dynamical features arising in the stochastic framework.

We propose a formal account of how structured, shared knowledge about social relationships shapes action interpretation. The model represents relationships as constraints in a social environment, analogous to boundaries or obstacles in a physical environment and operating within the same generative model, but exerting distinct constraints on action. As an initial test of this framework, we draw on research across the social sciences to capture in the models how one dimension of relationships – formality versus intimacy – shapes how people interpret interpersonally vulnerable behavior. We test this account in stories of naturalistic everyday situations, extending structured models of action understanding to open-ended contexts. Across six preregistered experiments (N = 1,554), the model captures people's inferences about desires, physical environments, and social relationships. This work formalizes how relationships can constrain – and be revealed through – everyday action.

Identifying patients with similar clinical characteristics is important for cohort discovery, clinical decision support, and personalized medicine. However, it is unclear whether learned patient representations preserve the similarities present in the original clinical data. We propose a correspondence score to directly assess how well these relationships are preserved, independently of labeled outcomes. The score combines complementary relationship-preservation metrics, with dataset-specific weights estimated through Bayesian optimization. We evaluated the approach on four complete blood count datasets using representations generated by five tabular foundation models and six classical transformation methods. Correspondence differed substantially across methods and representation dimensionalities. Patient relationships were generally best preserved at intermediate dimensionalities, while stronger reduction or expansion introduced greater deviations from the original data structure. Foundation models preserved correspondence more consistently than classical methods, with improvements ranging from approximately 5% to over 125%. The proposed score provides a way to evaluate representation quality before downstream use and supports informed selection of representation methods and dimensionalities.

Protein language models (pLMs) encode information about protein sequences which enable downstream tasks such as structure prediction, but their internal representations are not well understood. Sparse autoencoders (SAEs) provide a promising tool to disentangle latent pLM representations into interpretable features, but existing annotation pipelines largely rely on protein-level annotations derived from database labels and LLM annotations of top activating sequences. Such annotations can overlook the localized residue-level and geometric patterns encoded by sparse features. We introduce an automated and scalable method for interpreting SAE features in ESM-2 by using geometrically inspired features of the protein $\text{C}_\alpha$ backbone. Across ESM-2 8M layers, an FDR-controlled discovery analysis shows that local geometry is significantly associated with many SAE features, with varying levels of predictive strength, expanding coverage beyond database and sequence-based methods. In particular, geometry can distinguish SAE features sharing the same database annotation, revealing substructure within known biological labels. A significant portion of SAE features activate on unannotated metagenomic protein sequences enabling us to use our SAE annotations to better understand these sequences. In addition, ablation experiments at the level of contact prediction show that removing found geometric features shifts ESM-2's predicted contact maps in the direction of the descriptor. This provides a robust method of annotating proteins activated within SAE neurons at a residue level, providing a bridge between mechanistic interpretability and structural biology.