Papers

We study whether the loss design of High-Fidelity Generative Image Compression (HiFiC), a GAN-based neural codec originally built for natural photographs, can be adapted to preserve biologically meaningful structure in Hi-C chromatin contact maps under lossy compression. Standard image compression, including HiFiC in its original form, optimizes for human visual perception; but a Hi-C contact map is normally distributed together with its numeric matrix file (.cool/.mcool), which downstream genomic analysis tools consume directly. Aggressive compression that looks acceptable to the eye can nonetheless blur or delete loops and topologically associating domain (TAD) boundaries that these tools depend on. We modify HiFiC's distortion term with a spatially-weighted MSE that up-weights biologically salient regions (loops, TAD boundaries, stripes, compartment structure) and add an insulation-score loss term that directly penalizes loss of TAD boundary sharpness. We describe a three-phase fine-tuning strategy that adapts a pretrained HiFiC checkpoint to the Hi-C domain without catastrophic forgetting. We evaluate the resulting system, HiFiC-G, using both conventional image-quality metrics (PSNR, SSIM) and genomics-domain preservation metrics (loop/TAD/compartment/stripe preservation percentage) across two cell lines. HiFiC-G preserves local structure, meaning stripes and TAD boundaries, substantially better than the metrics alone would suggest, while long-range A/B compartment structure remains poorly preserved; we show this gap tracks genomic scale and is consistent with a specific architectural cause, the fixed-size tiling that both HiFiC-G and the original HiFiC rely on for memory efficiency.

This paper develops a phenomenology-first approach to artificial consciousness by reframing consciousness as the subjective experience enacted through an agent's interface with the world. We shift the methodological focus to first-person structures, modeled mathematically by categories derived from Q-networks to capture actions and phenomenological invariants. In this framework, Q-networks are conceptualized as relational interfaces encoding agent-world interaction, analogous to how the dynamical states of a computer depend on its sensory inputs, previous states, and actions. Our work provides a rigorous framework for interface consciousness to describe computational systems that embed information-processing into phenomenological structure. The approach aligns with 4E approaches to cognition by emphasizing enactive, embedded, and extended dimensions of experience. The paper thus offers a principled, relational, and phenomenological account of artificial phenomenology grounded in categorical mathematics.

Alzheimer's disease (AD) progresses as a continuous biological process, whereas most existing neuroimaging-based artificial intelligence methods remain limited to discrete diagnosis or clinical score prediction from cross-sectional imaging. In this work, we propose Disease Continuum Positioning (DCP), a longitudinal Bayesian Learning framework that continuously estimates disease severity from longitudinal diffusion tensor imaging (DTI). Specifically, DCP models disease severity as a low-dimensional probabilistic latent variable by jointly integrating longitudinal observations with weak clinical supervision, from which the proposed Disease Continuum Score (DCS) is derived to quantify an individual's position along the Alzheimer's disease continuum together with its associated uncertainty. Extensive experiments on the Alzheimer's Disease Neuroimaging Initiative (ADNI) cohort demonstrate that DCP consistently outperforms representative disease progression methods. More importantly, comprehensive validation analyses show that DCS accurately characterizes disease severity, exhibits strong clinical relevance, preserves longitudinal disease evolution, and predicts future disease conversion. These results suggest that DCS provides a quantitative imaging-derived representation for continuous assessment of Alzheimer's disease progression beyond conventional diagnostic labels and clinical scores.

This study aims to explore the applicability of Transformer-based models for genetic sequence classification by evaluating their performance in predicting hepatitis C virus (HCV) genotypes and subtypes after fine-tuning. A total of 2,881 HCV whole-genome sequences obtained from the Los Alamos HCV Sequence Database were used, including genotypes 1 to 6 and all confirmed subtypes. Genotypes 7 and 8 were excluded due to an insufficient number of samples. The fine-tuning process was based on several datasets that differed in fragmentation method, data volume per file, and labeling. In genotype classification, fine-tuning strategies employing homogeneous fragmentation and balanced sample distribution resulted in higher performance, with precision ranging from 98.48% to 100%. In contrast, fine-tuning conducted using a fragmentation strategy that caused data imbalance, along with an arbitrary distribution of samples across training files, achieved a precision of 48.12%, which is considered low compared with other models. This configuration, which was also manually evaluated, resulted in a high error rate in genotype 5 prediction due to its low frequency in the datasets used. In subtype classification, the best-performing fine-tuning approach achieved 99.89% accuracy and 99.87% precision. Models that included additional genotypes showed a slight decrease in performance due to the increased complexity of the task. This study demonstrates that, when fine-tuning datasets contain properly fragmented, distributed, and labeled genetic sequences, Transformer-based neural networks can achieve high performance and are a promising approach for HCV genotype and subtype classification.

Lung cancer screening with low-dose chest computed tomography reduces mortality, but its impact is limited by uptake, adherence, and management challenges. Blood-based cell-free DNA (cfDNA) biomarkers offer a complementary approach, although early detection remains difficult because of lung cancer heterogeneity and high-dimensional, nonlinear molecular signals. We evaluated quantum-classical hybrid machine learning for lung cancer detection using DNA fragmentomics and DNA methylation. After feature selection, models were trained using 20- and 40-feature subsets. Features were encoded into quantum Hilbert space using angle and dense-angle feature maps with multiple entanglement strategies. Fidelity-based quantum kernels were computed with exact statevector simulation and integrated with precomputed-kernel SVM and kernel-PCA logistic regression and compared with an SVM model trained on the original features. This framework enabled systematic evaluation of how encoding and entanglement design affect classification. Across repeated held-out evaluations, quantum-kernel models achieved competitive performance on both datasets. For fragmentomics, several 20-feature configurations improved AUC relative to a classical SVM baseline, suggesting effective capture of nonlinear cfDNA fragmentation structure. For methylation, the classical SVM achieved the highest AUC, although selected quantum models remained competitive and improved specificity in some cases. Increasing features from 20 to 40 did not consistently improve performance and often increased variability. Overall, these results support quantum kernel methods as a promising approach for cfDNA-based lung cancer detection.

Boolean networks provide a compact framework for modeling regulatory systems, yet their rapidly expanding state spaces make systematic dynamical analysis challenging. Here, we systematically enumerate all non-isomorphic two-node signed regulatory subgraphs with their admissible Boolean update rules and exhaustively characterize their state-transition graphs (STGs) under both synchronous and asynchronous updating. By extracting quantitative dynamical descriptors, we move beyond static wiring diagrams toward a behavior-based classification of network structures. The synchronous analysis identifies deterministic fixed-point and cyclic attractor regimes, while the asynchronous analysis captures one-node-at-a-time dynamics through terminal strongly connected components of the asynchronous STGs. Comparing the two update schemes distinguishes dynamical behaviors that are robust across update assumptions from those that depend on synchronous updating. Despite the diversity of graph-rule combinations, many realizations converge to a limited repertoire of dynamical classes, revealing substantial redundancy between structural and dynamical representations. We further propose a testable framework for investigating whether such local dynamical classes can be used to characterize selected subnetworks within larger Boolean models. To facilitate exploration and reproducibility, we developed BORNA, an interactive web application for exploring the complete network catalogue, STGs, and synchronous and asynchronous simulations, available at https://jafarilab.github.io/BORNA/. Together, these results provide a systematic reference for minimal Boolean network dynamics and a foundation for extending behavior-based analysis to larger regulatory systems.

In recent decades, the early olfactory system (EOS) of the fruit fly has become a leading model for studying olfactory processing and associative memory, owing in part to a well-characterized feedforward pathway that feeds the processes underlying associative memory and by examining the role played by a handful of neurons and synapses. The recent completion of dense electron-microscopy connectomes provides high quality visualizations of every cell type, neuron, and synapse along the early olfactory pathway. Yet a wiring diagram, however complete, does not by itself reveal the functional logic of a neural circuit. Reviewing the EOS connectome and synaptome datasets of the past fifteen years, we note that the feedforward pathway is embedded in dense local feedback circuits of large scale multi-input multi-output neurons. A systematic understanding of feedback loop abstractions, and their capacity to govern the input/output transformations at each neuropil stage, is the underlying foundation of the functional logic of the early olfactory circuits. In addition, we argue that a quantitative account of the functional logic requires an explicit model of the odorants present in the natural environment. Consisting of odorant objects, such a model defines the semantics and syntax of olfactory information processing, and calls for new distance measures for classifying the odorant semantics in support of associative memory operations. Furthermore, odor information processing must abide by causality, treating the circuit as a real-time, stage-by-stage cascade of giant local feedback loops.

Responses in the sensory thalamic nuclei are modulated by perceptual tasks. Whether such response modulations rely on feedback from cerebral cortex in humans is unknown. Here, we addressed this question in the context of visual speech recognition: the visual sensory thalamus, i.e. the lateral geniculate nucleus (LGN), has differential BOLD-responses to visual speech than non-speech control tasks. We tested whether such response modulation relies on the function of the visual association cortex, specifically visual-motion area V5/MT. We applied inhibitory transcranial magnetic stimulation (TMS) over bilateral visual-motion sensitive areas V5/MT on 26 healthy adults. Subsequently, participants performed a visual speech and a colour recognition task on identical muted videos of speaking faces during functional magnetic resonance imaging (fMRI). The LGN showed a significant signal change between the visual speech task and the colour task following Vertex stimulation as active control region. This modulation was significantly reduced following inhibitory V5/MT stimulation. V5/MT stimulation also reduced task-dependent functional connectivity between V5/MT and the LGN. These results identify corticothalamic feedback as integral mechanism in visual processing. In particular, the visual association cortex has a causal role in modulating LGN responses during speech recognition.

Bioassay activity prediction is often data-limited because drug-discovery datasets rely on time-consuming and expensive wet-lab experiments for data generation and evaluation. This challenge has inspired recent research into molecular foundation models (MFMs), which aim to encode general-purpose chemical knowledge into molecular representations that generalize well in data-constrained scenarios. This paper presents Monroe, a new MFM with several innovations over the existing state of the art: increased scale allowing pre-training on over 81 million molecules from the PM6 quantum chemistry dataset; improved graph representation of stereochemistry; improved training losses including conformer denoising and embedding decorrelation; improved multi-task learning; and the use of a prior-data-fitted model (TabPFN) for downstream in-context prediction. Our evaluations use a principled pairwise comparison framework that measures statistically significant performance differences. Across established Polaris benchmarks, Monroe matches or exceeds existing MFMs, while on activity cliff benchmarks, designed to assess utility for molecular discovery, it achieves significant improvements over prior methods. Finally, ablation and transfer experiments show that PFN-based downstream predictors also substantially improve two leading existing models, MiniMol and CheMeleon, yielding new state-of-the-art variants we call MiniMol_PFN and CheMeleon_PFN, suggesting that our downstream adaptation strategy generalizes beyond Monroe. Source code is at github.com/blazejba/monroe.

The tensile properties of single collagen fibrils, the building block of load-bearing tissues, have been studied extensively by nanomechanical techniques and molecular dynamics simulation. However, the deformation pathway of collagen molecules within fibrils has not yet been observed experimentally. In addition, the role played by divalent and trivalent enzymatic crosslinks in modulating this deformation pathway is poorly understood. Here we used polarization controlled second harmonic generation (SHG) imaging combined with atomic force microscopy (AFM) to characterize the molecular state of collagen triple helices within stretched single collagen fibrils. The fibrils were extracted from a pair of bovine leg tendons from the same animal in order to compare fibrils with a high amount of immature divalent crosslinks to fibrils with a high amount of mature trivalent crosslinks. By selecting fibrils with a large SHG intensity gradient along their length and then imaging the same fibrils by AFM we were able to link the observed intensity gradient with a gradient in D-band strain and a gradient in molecular strain as estimated from the SHG anisotropy parameter rho. In contrast to previous studies at the tendon scale, we observed that the SHG molecular strain is always larger than the D-band strain for all fibrils with this difference being largest for fibrils rich in divalent crosslinks. By analyzing the behavior of the relative density of SHG emitters as a function of molecular strain, we observe a two-state transition from an SHG producing to a non SHG producing state with a free energy barrier between 6 and 10 kBT that we propose corresponds to the local untwisting of the collagen triple helix superhelical twist which likely preceded bond rupture and loss of the SHG signal. We also show that trivalent crosslinks tend to delay the transition onset compared to divalent crosslinks.

Simultaneously engineering multiple protein properties remains a major challenge. Existing machine learning-based pipelines for protein engineering often model properties separately, failing to capture their dependencies and trade-offs. Here, we systematically evaluate how Bayesian parameterization on Multitask Neural Networks can enable robust simultaneous protein engineering under scarce, noisy experimental data. We curated a comprehensive set of 27 multiparameter protein datasets. Then, we compared three algorithm architectures spanning low to full Bayesian parameterization across 16 sequence representations and dimensionality reduction (2,592 models). Bayesian Last Layer models delivered the strongest overall accuracy, generalization, and calibration, ranking as the top-performing model on 70% of benchmark datasets. Dimensionality reduction improved predictive performance by up to 42% and enhanced calibration up to 57% across architectures. Notably, simple One-Hot encoding achieved top performance on 25% of benchmark datasets, particularly with larger datasets. These results establish practical design principles for reliable and data-efficient multiparameter protein engineering.

Spatial aliasing occurs when two or more distinct locations produce highly similar place-cell representations, primarily due to environmental symmetry or repetitive structures. This issue is most pronounced when place representations are constructed solely from boundary vector cell (BVC) inputs, because symmetric or repetitive structures can yield indistinguishable sensory patterns across multiple locations in an environment. This work introduces grid cell signals to mitigate spatial aliasing in such settings. Because grid cells contribute periodic, internally generated spatial signals that vary independently of environmental geometry, they play a key role in disambiguating perceptually identical locations. We integrate multiple modules of analytically constructed grid cells with BVC-driven place cells and show that this leads to a 94–99% reduction in spatial aliasing relative to a BVC-only baseline across three environments: an open environment without obstacles; an environment with a cross-shaped central obstacle creating high visual symmetry; and a maze environment. The greatest improvement occurs in the environment with the highest visual symmetry. These results indicate that grid cells provide information complementary to boundary-based inputs, yielding more reliable place representations in geometrically ambiguous environments.

A fundamental problem in the algebraic study of biochemical reaction networks is to characterize degeneracy. Two-dimensional zero-one networks, where each reactant appears with stoichiometric coefficient zero or one, constitute the smallest biologically relevant class capable of exhibiting degeneracy. In this paper, we provide a complete classification: a two-dimensional zero-one network without trivial species is degenerate if and only if it is a consistent subnetwork of a species refinement of one of two prototypical networks, namely the complete paired-exchange network or the catalytic-pair conversion network. Equivalently, in matrix-theoretic terms, degeneracy is determined entirely by the row patterns of the stoichiometric and reactant matrices, and can be verified by purely structural inspection without computation. Remarkably, every such degenerate network has a steady-state system consisting entirely of binomials, so its steady-state variety is toric. These results also bear on absolute concentration robustness: since any network exhibiting this property necessarily contains a degenerate subnetwork, the minimal degenerate networks characterized here serve as fundamental building blocks for constructing such networks.

Background and Objective: Reliable atrial fibrillation (AF) detection from electrocardiogram (ECG) signals remains challenging in real-world clinical settings due to variable lead configurations, cross-dataset domain shifts, and pervasive physiological and technical artifacts. So we develop a robust and generalizable deep learning model for accurate AF detection.\\ Methods: We propose the Dual-Codebook Graph Collaborative Network (DCGCNet), a novel end-to-end vector-quantized variational autoencoder that jointly performs AF classification and ECG reconstruction. DCGCNet introduces two key components: (1) a Local-Global Contrastive Module for learning noise-invariant representations, and (2) an Adaptive Codebook Vector Quantizer that dynamically refines codebook prototypes to better align with input data distributions, thereby preventing codebook collapse and enhancing generalization.\\ Results: DCGCNet achieves state-of-the-art performance in standard intra-dataset 12-lead evaluation and demonstrates exceptional cross-dataset generalization across seven diverse settings, consistently attaining AUC > 0.98 in all cases. Furthermore, it maintains high diagnostic accuracy under realistic noisy conditions, including baseline wander, powerline interference, and EMG artifacts.\\ Conclusions: DCGCNet establishes a new benchmark for robust, generalizable, and noise-resilient AF detection, showing strong potential for deployment in real-world clinical environments.

We introduce Malaria-Instruct, a curated instruction-following dataset derived from the ChEMBL Legacy Malaria corpus for Malaria virtual screening, and conduct a systematic evaluation of five open-source LLMs; Gemma-2 2B/9B, TxGemma-2B/9B, and LlaSMol-Mistral-7B, on a rigorous out-of-distribution data split. Performance was benchmarked against classical ML models (Random Forest, XGBoost) and frontier proprietary models (Gemini 2.5, OpenAI o3) under few-shot conditions. Fine-tuned LLMs substantially outperformed all baselines: TxGemma-9B achieved the highest ROC-AUC ($0.731 \pm 0.005$) and LlaSMol-Mistral-7B the best enrichment factor (EF@1% $\approx$ 4.99). Domain-specific fine-tuning proved categorically indispensable with TxGemma-9B collapsing from ROC-AUC 0.731 to 0.499, under its best few-shot condition, and neither Gemini 2.5 (ROC-AUC $\approx$ 0.53) nor o3 (ROC-AUC $\approx$ 0.59) achieved reliable discrimination without fine-tuning. Biomedical pretraining conferred a measurable advantage at equivalent scale, while chemistry-aware pretraining yielded superior prospective enrichment. Fine-tuned open-source LLMs represent a compelling, resource-efficient paradigm for antimalarial VS, outperforming both classical pipelines and proprietary reasoning models under structurally challenging conditions.

Adaptive therapy has emerged as a promising treatment strategy that exploits within-tumour competition to delay disease progression. Implementation, however, typically relies on indirect measurements of tumour burden and must account for potentially substantial patient heterogeneity. In this work, we capture patient-to-patient variability, parameter uncertainty, and imperfect biomarker measurements with a mathematical and statistical model that we calibrate to clinical prostate cancer data using a Bayesian inference framework. We use the resulting virtual cohort to demonstrate that, within the simple but now well-established Lotka-Volterra-based model, adaptive therapy robustly improves time-to-progression for the subset of patients that are predicted to eventually progress by the model. To account for other risk factors associated with larger tumour volumes, we introduce a new metric based on the risk of metastasis that demonstrates how adaptive therapy may be disadvantageous when sustained tumour burden is also considered. Given the ubiquity of uncertainty in oncology, we then describe several future modelling directions that also capture uncertainty in the temporal evolution of the underlying tumour or biomarker dynamics. Finally, we demonstrate how model misspecification and non-identifiability can lead to unreliable predictions, especially if uncertainty is inadequately captured.

Emergent systems - systems with multiscale interactions - arise through iteration than forward design, leaving principles for building them under-explored. Current artificial intelligence architectures, while useful for generation, provide no logic for construction. Inspired by statistical phylogenetics, we show that inferring scales of system entropy then constraining information from low to high entropic scales yields stepwise constructive models. We term this architecture the Layer-Restricted Boltzmann Machine (LRBM). Applied identically to digit images, natural language, and enzyme sequences, LRBMs follow a common hierarchical logic: lower layers encode global structure, higher layers fine-grained features. As a stringent test, we assayed 160 synthetic chorismate mutase (CM) constructive trajectories alongside 1,130 natural homologs in vivo. Designed CMs functioned up to 53% divergent from nearest natural homolog. Fold emerged at intermediate layers while function required all layers, illustrating that fold was necessary but insufficient for function. Our results demonstrate a shared, learnable logic for constructing emergent systems.

Genomic analysis on viruses such as SARS-CoV-2 variants: Beta, Gamma, Delta, and Omicron is heavily dominated by classical bioinformatics methods, including Sequence Alignment, Phylogenetic Analysis, and Mutation Frequency Statistics. These approaches use pairwise codon or nucleotide distance matrices to analyze gene sequences, treating them as linear strings rather than capturing their complex contextual interdependencies. We proposed GenEx, a pipeline that converts raw gene sequences into codon co-occurrence graphs and extracts more than 25 graph features. Our two most prominent techniques for graph generation and feature extraction are MSCG (Multi-Scale Codon Co-occurrence Graph) and LAPCG (Linear-time Adjacency PMI Codon Graph). Using these algorithms, we treated codon sequences as structured symbolic vocabularies interpretable to codon co-occurrence graph analysis, a representational paradigm borrowed from computational linguistics. Another major contribution includes implementing a spectral graph feature extraction using Singular Value Decomposition (SVD), using the squared singular value ($\sigma^2$) instead of the traditionally used eigenvalue, which helped us to amplify the separation between dominant and subdominant spectral components, thereby enhancing inter-class separability in downstream classification. And to further demonstrate that our method works, we trained 23 benchmarked ML models against the latest SARS-CoV-2 variants, achieving remarkable results in detecting all SARS-CoV-2 variants.

Chimeric antigen receptor (CAR) T-cell therapy has transformed the treatment of B-cell acute lymphoblastic leukaemia (B-ALL). Despite high initial response rates, a substantial fraction of patients relapse, often due to loss of CAR T-cell persistence, antigen escape, or immune-privileged sites that shield tumour cells. Prolonged CAR T-cell persistence is clinically associated with durable remission, but why it is required remains poorly understood. To address this, we develop and analyse the BEAM (Blast, Effector, Activated, Memory) model of CAR T-cell dynamics in B-ALL. BEAM extends predator–prey models with three CAR T-cell states (memory, activated, effector) coupled to a logistic growth equation for the blasts, calibrated against the FELIX trial of obecabtagene autoleucel in adult B-ALL. We find that both memory and effector persistence prevent relapse, but for distinct reasons: memory persistence sustains surveillance against low-burden or slowly proliferating residual disease, while effector persistence clears isolated blasts emerging from immune-privileged sites. The model further predicts a trade-off between immediate cytotoxicity and durable surveillance, and identifies initial tumour burden as a key modifiable factor for reducing antigen-negative relapse. Together, these results offer a framework for designing more durable, individually tailored CAR T-cell therapies.

Single-particle cryo-electron microscopy images a macromolecule as many noisy tomographic projections of its electrostatic potential. We reconstruct the protein backbone directly from such projections, as an atomic point cloud, without the intermediate step of reconstructing the 3D electrostatic potential map. We formulate this as an indirect shape matching problem: a point-cloud template of the backbone is deformed until its simulated projections agree with the data, with the structure observed only through the imaging operator. The deformation is computed via a gradient flow on a Lie group, and we derive the resulting framework in a general geometric setting before adapting it for single-particle cryo-electron microscopy. On synthetic data, we recover single- and multichain proteins and capture conformational transitions.

Preterm birth remains a major cause of neonatal morbidity and mortality worldwide. Electrohysterography (EHG), a noninvasive measure of uterine myoelectrical activity, has been studied for preterm-birth prediction, but performance estimates may be biased when segments from the same maternal record are split across training and validation data. We formalized the distinction between segment-level and patient-independent record-grouped validation, established a patient-independent benchmark on the Term-Preterm Electrohysterogram Database, and evaluated class-conditional conformal selective prediction. All 300 records (38 preterm) were analyzed across three prespecified regimes. A 92-feature elastic-net logistic model was evaluated with record-grouped nested cross-validation; preprocessing, model fitting, Platt calibration, and conformal estimation were confined to training data, and performance was calculated from strictly out-of-fold record-level predictions with 1,000 bootstrap resamples. Under patient-independent evaluation, AUROC was 0.493 (95% CI 0.467-0.520), AUPRC 0.122 (0.095-0.152), and Brier score 0.115 (0.094-0.137). AUROC was 0.514 at or before 26 weeks and 0.469 thereafter. At miscoverage alpha=0.10, marginal coverage was 0.897, abstention 72.7%, and singleton-prediction accuracy 0.624. These findings establish a record-separated reference benchmark for EHG prediction and a broader validation principle for segmented physiological data: the unit of resampling should correspond to the unit at which predictive performance is intended to generalize. Conformal prediction further quantifies when the available information supports a singleton classification and when uncertainty warrants deferral.

In the transcriptional networks of Escherichia coli and Saccharomyces cerevisiae, the eight feed-forward loop (FFL) motifs occur at markedly different frequencies. Although previous studies have linked the abundant C1- and I1-FFLs to specific dynamical functions, a common account of the broader pattern is lacking. To address this gap, we adopt an information-theoretic approach. An FFL transmits information through two paths that share an input and converge on an output, suggesting that their interaction may contribute to differences among motifs. To investigate this, we decompose input-output mutual information (MI) into pathway MI and interference MI (IMI). Here, the pathway MI collects the contributions of individual pathways, whereas IMI identifies the contribution that arises from their interference. We find that within a representative parameter regime, IMI values of FFLs produce a hierarchy that closely resembles the abundance ordering in E. coli, whereas total and pathway MIs do not. Motivated by this observation, we perform constrained optimization that identifies parameter regimes compatible with abundance-like targets for both S. cerevisiae and E. coli. We relate the optimized IMI hierarchy to pathway interference strength and local pathway sensitivities to understand the underlying biophysical origin. Our framework thus probes the correspondence between pathway interference, information transmission, and motif-abundance hierarchies.

Preterm birth (PTB) remains a major global health problem, and reliable non-invasive risk assessment remains difficult. Electrohysterography (EHG) records uterine electrical activity from the maternal abdomen and may support PTB assessment, but performance can be inflated when segments from the same recording are split across training and validation folds. We evaluated empirical mode decomposition (EMD) for term-versus-preterm classification using 26 pregnancy recordings (13 preterm, 13 term) from the public TPEHGT dataset. Annotated intervals and non-overlapping fixed 3-minute windows were compared, and the first four intrinsic mode functions (IMFs) were evaluated. Fourteen features from each of three EHG channels were assessed with nine classifiers using repeated five-fold recording-grouped cross-validation and recording-level aggregation. IMF1 gave the strongest mean performance. With fixed 3-minute IMF1 features, Random Forest achieved mean accuracy 0.8308, F1 0.7969, balanced accuracy 0.8308, MCC 0.6998, ROC-AUC 0.8157, and average precision 0.8877. IMF1 also outperformed matched filtered time-domain features across all reported mean metrics. Preterm recordings showed smaller, more regularly spaced peak-like events, lower temporal-energy measures, and higher entropy. These findings support further evaluation of IMF1-based EHG classification in larger independent cohorts.

Motivation: Low-dimensional embeddings are widely used to explore cell-state heterogeneity in single-cell and other high-dimensional biological data. Although many methods preserve local neighborhoods, they may distort the apparent sampling density of processed observations, altering the visual contrast between dense and sparse regions and complicating the interpretation of rare, transitional, or continuous cell-state populations. Results: We present DMT-Dens, a parametric manifold-visualization method built on a latent-token Transformer encoder. The model integrates rank-based manifold alignment with hard-pair aggregation. To preserve density, it optimizes a loss based on the Pearson correlation between k-nearest-neighbor log-radius estimates in the processed input and two-dimensional embedding spaces. Benchmark evaluations demonstrate strong density preservation, particularly on biological datasets, while retaining competitive label separability. Availability: Source code, data-processing scripts, and resolved experiment configurations are available at https://github.com/Ruizhe-wang/DMT-Dens.

Biomolecular design underpins applications from molecular recognition to therapeutics and synthetic biology, yet de novo interaction design remains challenging-especially for DNA/RNA, underexplored non-protein modalities with scarce, heterogeneous complex data and sharper geometric and chemical constraints. We introduce MCTH (Monte Carlo Tree Hallucination), an inference-only framework that casts all-atom sequence-structure co-design as uncertainty-aware planning over hallucinated states from pretrained folding and inverse-folding models, with optional biophysical control within the same decision loop. MCTH treats these models as frozen black-box operators and uses Monte Carlo Tree Search to allocate a fixed inference budget across competing design trajectories, incorporating model confidence and uncertainty, as well as cross-expert consensus/disagreement when multiple predictors are available. Across protein-RNA, protein-DNA, protein-protein, and protein-ligand design, matched-budget experiments show that adaptive search improves over simpler sampling and cycling strategies, while held-out AlphaFold3 and Chai-1 evaluations demonstrate transfer beyond the search-time oracle. MCTH provides a shared planning layer across modalities while allowing task-specific folding, inverse-folding, and biophysical modules, requiring no fine-tuning or backpropagation through component models.

Deep clustering models for single-cell RNA sequencing often assign cells through latent or centroid-based mechanisms that are difficult to inspect. We introduce scDNM-VAE (single-cell Dendritic Neuron Model Variational Autoencoder), a deep clustering framework that combines a variational autoencoder with a dendritic neuron-inspired head. Cluster assignments are governed by learnable signed synaptic weights and thresholds: the weight sign determines the direction of a gate's response to a latent coordinate, its magnitude controls steepness, and the weight-threshold pair determines the transition location. The trained clustering function can therefore be inspected directly without fitting a post-hoc explanation model. We benchmark scDNM-VAE on four datasets spanning immune, cortical, cardiac, and hematopoietic cells against scVI followed by KMeans and an MLP-DEC ablation. scDNM-VAE performs better than scVI on PBMC3k, comparably on the Human Heart Cell Atlas and Paul15, and worse on Zeisel, while producing biologically coherent marker-gene signatures. Ablating each cluster's three highest-magnitude synaptic dimensions causes numerically greater reassignment than random-dimension ablation across all datasets, but the margins are modest and negligible on Zeisel. These results show that signed dendritic gating supports competitive clustering with a parameter-inspectable decision function, while indicating that decision-relevant information is distributed across the latent space.

Colony demographic models often treat worker deaths as independent events, but shared environmental exposure may synchronize mortality within species. Using a 15-month field dataset of 18 Australian ant species and 106 cohorts, we quantified intra-specific mortality synchrony with the intra-class correlation coefficient (ICC) and tested body size and temperature as candidate correlates. Synchrony was widespread: credible intervals excluded zero in 9 of 18 species, and the assemblage mean ICC was 0.21 +/- 0.09, indicating that about 21% of daily mortality variance reflected shared within-species exposure. Body size did not buffer synchrony. Worker mass was unrelated to ICC (Spearman rho = +0.10, p = 0.69; WLS beta = +0.05, p = 0.35), decoupling how long workers live from when they die together. Temperature showed a suggestive lagged association: maximum temperature preceded mortality pulses by 1-7 days in 13 of 15 analyzable species, with mean peak r = 0.16, but no species survived Bonferroni correction. A sensitivity analysis using mean correlation across lags 1-7 days instead of the peak removed the assemblage-level signal (mean r = -0.03, p = 0.93), and ICC was unrelated to temperature-coupling strength (Spearman rho = -0.004, p = 0.99). Mortality synchrony is therefore widespread, size-independent, and only partly thermal. Species-specific thermal tolerance, rather than body mass, is a more promising candidate for explaining variation in synchronized mortality.

SARS-CoV-2 possesses a furin cleavage site (FCS) insertion in its spike protein, that is not found in other sarbecoviruses, and is responsible for its transmissibility in humans. The origin of the FCS insertion is unclear, and has been a point of debate regarding whether it arose naturally, or represents an artificial insertion. The FCS is destabilizing to spike protein, and as a consequence a compensatory mutation, D614G, rapidly emerged during the early pandemic and swept to fixation. Using a range of empirically determined parameters, a waiting time calculation can be used to determine the time interval between the appearance of the FCS and generation of the compensating D614G mutation. The waiting time is estimated to range from 37 to 65 days. Given the earliest detection of D614G on 1 January 2020, the latest dates for the appearance of the FCS range from October 28 to November 25 2019. This timeline is consistent with dates generated from phylogenetic considerations for the emergence of SARS-CoV-2 in late 2019, indicating that FCS appearance and virus emergence correspond. The waiting time approach enables discrimination between various scenarios of SARS-CoV-2 emergence, in particular cryptic circulation of a FCS-containing progenitor in the human population prior to late 2019 is unlikely, while prolonged circulation of a FCS-containing progenitor in bats appears inconsistent with absence of the compensatory mutation.

The emergence of organized spatiotemporal patterns is ubiquitous in oscillatory systems, from neural populations to engineered networks. Identifying these patterns and tracking how they evolve over time remains challenging, particularly when systems exhibit transient dynamics. Here, we introduce a framework based on spatial ordinal patterns to characterize the spatiotemporal dynamics of oscillatory systems. Our approach acts directly on the phase rather than the amplitude, with additional patterns introduced to account for near-equal phases. This symbolic representation encodes local spatial ordering relations, capturing both phase gradients and synchronized clusters within a single framework. From this construction, we define a spatial permutation entropy that quantifies the diversity of spatiotemporal patterns at each point in time, enabling the detection of transient dynamics and regime transitions as they occur. We show that this approach distinguishes phase-locked states with identical levels of global synchronization but distinct spatial organization and also characterizes partially synchronized states. We demonstrate the method on synthetic oscillator networks across multiple spatiotemporal regimes, and on resting-state EEG recordings from human volunteers, where it distinguishes different conditions within individual volunteers.

Competitive interactions can maintain diversity, yet coexistence is often fragile in well-mixed populations, where stochastic fluctuations can lead to extinction. This is the case in non-transitive systems, such as rock-paper-scissors dynamics, where no single type dominates globally. While spatial structure can stabilize these systems by providing refuges in space, it remains unclear whether analogous mechanisms can operate in time in well-mixed environments. Here, we develop a population-genetic framework showing that dormancy can act as a temporal refuge, preserving lineages and preventing collapse to fixation under interaction-driven fluctuations. We introduce a discrete-time Wright-Fisher model that combines generalized seed-banks with frequency-dependent interactions, allowing individuals to inherit their type from potential parents sampled across multiple past generations. This construction provides a tractable framework in which dormancy stores and later reintroduces lost types. In the case of either weak or moderate selection, we prove a multidimensional diffusion limit for the resulting type-frequency process and use it to analyze complex selective interactions. In non-transitive systems, dormancy stabilizes trajectories that would otherwise collapse through stochastic extinction, extends fixation times, and sustains coexistence. These effects cannot be explained solely by an increase in effective population size. Our results show that dormancy introduces temporal memory that qualitatively alters competitive dynamics, stabilizing otherwise fragile systems and enabling long-term coexistence.