PLOS Computational Biology: New Articles

Repurposing ritonavir to induce proteasomal degradation of IE1 for inhibition of human cytomegalovirus replication

2026-10-06T14:00:00Z

by Wanfeng Liu, Yu Shi, Xudong Guo, Chenyu Lou, Zhangyi Xie, Yitian Shen, Liang Zhang, Sunzhong Mao, Xiangyang Xue, Shiyu Feng

Human cytomegalovirus (HCMV) is a major pathogen affecting immunocompromised individuals. Current antiviral drugs target the mid-to-late stages of viral replication but are limited by toxicity and drug resistance. The IE1 protein serves as a key early regulator of HCMV replication; however, it has long been considered undruggable due to the lack of conventional drug-binding sites. In this study, by integrating molecular docking-based virtual screening with in-vitro surface plasmon resonance (SPR) validation, we discovered for the first time that the HIV protease inhibitor ritonavir may bind to the IE1 protein. Further experiments demonstrated that ritonavir significantly reduces IE1 protein levels in a time- and dose-dependent manner without affecting its mRNA expression. Mechanistic investigations revealed that ritonavir promotes selective degradation of IE1 through activation of the ubiquitin-proteasome pathway. Functional assays confirmed that ritonavir effectively inhibits HCMV replication, markedly reducing viral DNA loads and the production of infectious viral particles. Notably, ritonavir exhibited synergistic antiviral effects when combined with the first-line drug ganciclovir (GCV). In summary, this study successfully repurposes the approved drug ritonavir as a regulator of IE1 protein expression, demonstrating not only its anti-HCMV activity and synergistic effects with GCV but also providing a new therapeutic strategy for HCMV infection.

Spatial Transcriptomics As Rasterized Image Tensors (STARIT) characterizes cell states with subcellular molecular heterogeneity

2026-10-06T14:00:00Z

by Dee Velazquez, Caleb Hallinan, Roujin An, Kalen Clifton, Jean Fan

Imaging-based spatially resolved transcriptomics (imSRT) technologies provide high-throughput molecular-resolution spatial characterization of genes within cells. Conventional analysis methods to identify cell-types and states in imSRT data rely on gene count matrices derived from tallying the number of mRNA molecules detected for each gene per segmented cell, thereby overlooking subcellular heterogeneity that can be useful in defining cell states. To take advantage of the molecular-resolution information in imSRT data and potentially identify cell-states based on subcellular heterogeneity, we developed STARIT (Spatial Transcriptomics As Rasterized Image Tensors). STARIT converts transcripts within segmented cells in imSRT data into an image-based tensor representation that can be combined with deep learning computer vision models for downstream analysis. Using simulated and real imSRT data, we demonstrate that STARIT distinguishes transcriptionally distinct cell-types and further separates cell states based on subcellular transcript localization, which conventional gene count analysis fails to capture. By providing a standardized framework to encode subcellular molecular information in imSRT data, STARIT will enable deeper insights into subcellular heterogeneity and enhance the identification and characterization of cell-types and states that are overlooked by gene count representations.

Mobility hinders malaria elimination goals in the Brazilian Amazon

2026-10-06T14:00:00Z

by Nicholas J. Arisco, Pablo S. Fontoura, Cassio Peterka, Marcia C. Castro

Introduction

Human mobility across the Brazilian Amazon—shaped by frontier development, land-use change, mining activity, and climatic shocks—remains a major barrier to malaria elimination. Brazil aims to eliminate Plasmodium falciparum (Pf) by 2030 and Plasmodium vivax (Pv) by 2035, yet malaria incidence has plateaued since 2020. We quantify parasite-specific mobility and assess how importations drive malaria reintroduction and reestablishment in areas approaching elimination.

Methods

We analyzed 5,521,411 de-identified malaria cases from Sivep-Malaria (2003–2023) across 808 municipalities in the nine-state Brazilian Amazon. We (i) used multivariable monthly sequence analysis to classify “Municipal Malaria Importation Trajectories” (MMITs) using local/imported composition, incidence per 1,000, and source/sink network typologies; (ii) characterized mobility networks by parasite species and locality type (rural, urban, settlement, indigenous, mining); and (iii) developed a parasite-specific rule-based classification procedure to identify reintroduction, and defined reestablishment as sustained local transmission.

Results

Sequence analysis produced five MMIT clusters, spanning municipalities nearing elimination to areas with sustained high local transmission and distinct profiles of declining transmission with intense vs. moderate importation. Overall, 959,281 cases (17.38%) were imported. Cases were predominantly Pv (81.68%), with Pf comprising 17.15%; Pf was more common among imported than locally acquired infections. Mobility patterns shifted after mid-2020: by 2023, > 50% of exported cases originated in mining localities despite ~0.2% land coverage, with Pf importation increasingly concentrated in municipalities overlapping the Yanomami Indigenous Land. From 2006–2023, we identified 63 Pf and 221 Pv reintroduced locally acquired cases following ≥3 years of zero local transmission, leading to reestablishment in 6 and 23 municipalities, respectively, and generating 864 subsequent Pf and 6,524 subsequent Pv cases. Post-reintroduction transmission persisted for long periods (mean 762 days for Pf; 815 days for Pv).

Conclusion

Malaria elimination in Brazil is increasingly hindered by parasite-specific mobility, especially linked to mining–indigenous confluences, and by repeated reintroduction in receptive municipalities nearing elimination (notably along the arc of deforestation). Mobility-informed surveillance and targeted reactive strategies are essential to achieve and sustain Brazil’s malaria elimination goals.

HiFIseek: Gene-specific enrichment of high-impact mutations in associated genomic regions

2026-10-06T14:00:00Z

by Rafael Riudavets Puig, Denise G. O’Mahony, Ina Skaara Brorson, Ksenia Sokolova, Vessela N. Kristensen, Anthony Mathelier

Transcription is regulated through the sequence-specific binding of transcription factors (TFs) to cis-regulatory regions (CRRs); DNA variants in CRRs can disrupt TF-DNA interactions and gene regulatory networks through a cascading effect. Despite a few cases of reported recurrent cis-regulatory variants (CRVs), another approach for detecting non-coding functional events rely on combining gene-CRR associations capturing the regulatory elements controlling gene transcription, and functional impact (FI) scores quantifying the deleterious effect of variants. While many methods exist for each, how much they agree and how strongly the choice of CRR-FI score combination impacts the prediction of cis-regulatory altered genes has not been systematically assessed. Here, we demonstrate that gene-CRR associations and FI scoring methods exhibit little consensus, highlighting the impact of the choice of a specific CRR-FI score combination in a given analysis. These observations point to the need for a systematic approach to evaluate several gene-CRR and FI scoring methods in a unified framework. To address this gap, we developed HiFIseek, a Snakemake pipeline of two enrichment detection methods exploring the enrichment of high FI variants in CRRs associated with a gene using all region-score combinations. We applied HiFIseek to detect genes exhibiting high functional impact CRVs in eight cancer cohorts and a set of breast cancer risk-associated single-nucleotide polymorphisms (SNPs). We demonstrate that cancer driver genes exhibit a higher frequency of FI CRVs compared to non-cancer genes. In the cancer cohorts, HiFIseek detected genes pointing to cancer-related pathways with enrichment for high FI CRVs despite a small degree of agreement across CRR-FI score combinations. In the SNP cohort, HiFIseek found again small consensus across CRR-score combinations, one of which returned 18 known cancer-related genes, including BRCA1, showing an enrichment of high-impact CRVs.

CardamomOT: A mechanistic optimal transport-based framework for gene regulatory network inference, trajectory reconstruction and generative modeling

2026-10-06T14:00:00Z

by Yann Maugé, Elias Ventre

A key challenge in inferring gene regulatory networks (GRNs) governing cellular processes such as differentiation and reprogramming from experimental data lies in the impossibility of directly measuring protein dynamics at the single-cell level, which prevents establishing causal relationships between regulator activity and target responses. In earlier work, we introduced CARDAMOM, an algorithm that uses temporal snapshots of scRNA-seq data to calibrate a GRN-driven mechanistic model of gene expression. However, this method had several limitations: it could only rely on the relative ordering of time points rather than their exact labels, imposed restrictive quasi-stationary assumptions on protein dynamics, and depended on multiple hyperparameters. Here, we present CardamomOT, a new method based on the same mechanistic model that jointly reconstructs the GRN and unobserved protein trajectories from the data within a mechanistic optimal transport framework. By incorporating exact time labels and priors on protein kinetic rates from the literature, and substantially reducing the number of required hyperparameters, our approach addresses these limitations and substantially improves the accuracy and robustness of GRN calibration. We validate our framework on both in silico and experimental datasets, demonstrating computational scalability and consistently improved performance over state-of-the-art methods in both GRN and trajectory reconstruction on simulated datasets, and, on experimental datasets, reconstruction of cellular trajectories, velocity fields and latent protein levels that are mutually consistent, together with GRN structures consistent with known biology. We also show that these improvements make the calibrated mechanistic model suitable to be used as a generative model to generate testable predictions of cellular responses to unseen perturbations. To our knowledge, this is among the first methods to explicitly integrate mechanistic GRN inference, trajectory reconstruction, and simulation of realistic datasets into a unified framework for scRNA-seq time series analysis.

Omics data discovery agents: Agent-supported retrieval, reanalysis, and synthesis of published omics data

2026-10-06T14:00:00Z

by Alexandre Hutton, Jesse G. Meyer

The biomedical literature contains a vast collection of omics studies, yet most published data remain functionally inaccessible for computational reuse. When raw data are deposited in public repositories, essential information for reproducing reported results is dispersed across main text, supplementary files, and code repositories, and in the rarer cases where intermediate data (e.g., protein abundance files) are shared, their location is irregular. Here we present an agentic framework for the agent-supported retrieval, reanalysis, and synthesis of published omics data. The system employs large language model (LLM) agents with access to tools for fetching omics studies, extracting article metadata, identifying and downloading published data, executing containerized quantification pipelines, and synthesizing results across studies. Applied at corpus scale, the pipeline catalogued dataset references across thousands of PubMed Central articles; we report these as descriptive system outputs rather than as a validated measure of extraction accuracy. Using model context protocol (MCP) servers to expose containerized analysis tools, the agents retrieved and re-quantified data in five end-to-end reanalyses spanning data-dependent and data-independent proteomics and bulk RNA-seq. All five reanalyses completed, each with documented human guidance and workflow accommodations, and reproduced the authors’ deposited abundances with high per-sample correlation (0.85–0.997) and strongly concordant differentially expressed features (fold-change Spearman 0.88–0.91), with no direction reversals among features called differentially expressed in both analyses; residual differences in significant-feature lists were attributable to threshold placement, tool-version, and preprocessing differences rather than to the underlying quantities. We further demonstrate that agents can identify semantically similar studies, judge data compatibility, and synthesize findings across studies, including a random-effects meta-analysis that recovered consistent protein regulation in liver fibrosis. Rather than a validated benchmark of literature-wide performance, this work is a feasibility demonstration together with an auditable, reusable toolset, establishing a foundation for prospective evaluation of automated omics-data reuse.

Bioactivity-driven prediction of antibacterial synergy using machine learning models

2026-10-05T14:00:00Z

by Hannie Yousefabadi, Mahya Mehrmohamadi

Motivation

Combination therapies have demonstrated improved efficacy across diverse clinical settings. Predicting antibacterial drug synergy remains difficult due to strain variability and the limited scale of experimentally tested combinations. Existing machine-learning approaches for synergy prediction often rely on permissive cross-validation schemes that allow drug pairs to re-appear across folds, leading to inflated performance metrics. A rigorous evaluation framework and scalable feature representation are needed for robust generalization.

Results

We assembled a curated dataset of 3,160 drug–pair–strain interactions covering 97 antibacterial compounds and 10 bacterial strains. We then developed HALO (Held-out Antibacterial interaction Learning from latent bioactivity Observations), a synergy-prediction framework in which each drug pair is encoded using multi-level Chemical Checker (CC) similarity features spanning chemical, targets, networks, cellular, and clinical bioactivity domains. Using strictly nested, drug-pair heldout cross validation with fold-internal feature selection, HALO achieved consistent generalization to unseen combinations (ROC–AUC = 0.78). Performance dropped notably under increasingly stringent heldout schemes and inflated under random data splits, underscoring the importance of rigorous evaluation for this task. Despite these conservative conditions, HALO transferred to two independent datasets, achieving ROC–AUC = 0.90 and 0.71 and average precision = 0.76 and 0.94 for distinguishing synergy from antagonism. Together, these results demonstrate that multi-level bioactivity signatures provide a scalable, interpretable basis for predicting antibacterial relationships, while clarifying the realistic performance limits of current models under truly leakage-free evaluation.

DeCTCF: Decoding CTCF binding sequences by leveraging predicted epigenomic features

2026-10-05T14:00:00Z

by Lu Chai, Jie Gao, Tinghe Guo, Teer Ba, Zihan Li, Junjie Liu, Yong Wang, Lirong Zhang

CTCF is a key architectural protein with diverse roles in genome organization and gene regulation, yet how it achieves these roles in different contexts remains unclear. Pretrained sequence-based models such as Sei provide predicted epigenomic features that can be used in downstream analyses of regulatory elements. Here, we developed DeCTCF, an integrative computational framework that uses pretrained Sei predictions to analyze 236,552 CTCF binding sequences by integrating CTCF ChIP-seq data from 118 human cell lines. By leveraging predicted epigenomic features from the Sei model, we grouped these CTCF binding sites into 20 clusters. These clusters can be annotated into distinct functional modules, including a major module associated with 3D chromatin architecture and three lineage-associated modules. The lineage-associated modules reveal associations between candidate co-factors and CTCF’s context-dependent functions. For example, several clusters enriched in the three stem cell lines included in our dataset also showed enrichment of ZIC-family and were associated with gene sets related to pluripotency and neurodevelopment. We further observed associations between cluster-level CTCF ChIP-seq signal profiles and chromatin-loop annotations: single-peak profiles were reproducibly associated with higher loop interaction scores, whereas double- and triple-peak profiles showed distinct loop-pairing preferences. Overall, our study offers a systematic map of CTCF’s modular organization by leveraging predicted epigenomic features and reveals context-associated regulatory patterns that underlie its regulatory diversity.

Quantifying antibiotic susceptibility and inoculum effects using transient dynamics of Pseudomonas aeruginosa

2026-10-05T14:00:00Z

by Sarah Sundius, Jennifer Farrell, Kelly L. Eick, Rachel Kuske, Sam P. Brown

Antibiotics are a cornerstone of modern medicine, targeting pathogen cells by disrupting essential cellular processes. However, standard antibiotic susceptibility metrics (e.g., MIC) and textbook models neglect transient dynamics and density-dependent effects, despite their ubiquity in nature. In clinical infections, where bacterial populations are the units we treat, this can increase the risk of under treatment. To address this gap, we generate high resolution optical density time series data for Pseudomonas aeruginosa (3 antibiotics, 12 doses, 7 inoculum sizes, 4x replication), enabling gradient estimation and gradient-based model parameterization. We develop a dynamics-led computational pipeline that (1) evaluates population scale ordinary differential equation models in the context of estimated time derivative data, and (2) classifies transient dynamics in dose-inoculum space using unsupervised clustering. Applied to our data, the pipeline identifies an ordinary differential equation model with a saturating antibiotic-loss term and a threshold-dependent weak Allee term that recapitulates and quantifies classic rate, yield, and inoculum effects of antibiotics. In addition, our model and clustering approach suggest a set of novel metrics, defining thresholds separating distinct dynamical regimes. Beyond antibiotic data sets, our approach utilizing a derivative-based fitting algorithm and clustering of derivative trajectories is applicable to any biological time series with controlled perturbations and variable initial conditions.

A generalized scaling law reveals synchrony-driven reorganization of brain dynamics across human lifespan

2026-10-05T14:00:00Z

by Suman Saha, Chittaranjan Hens, Priyanka Chakraborty, Tomasz Kapitaniak, Gustavo Deco, Arpan Banerjee, Syamal Kumar Dana, Dipanjan Roy

Ecological systems obey universal scaling laws such as Taylor’s law, which links population variance to the sample mean. Whether other complex systems with integrated and segregated network topology, the human brain, in particular, follow analogous scaling laws, and how these laws behave under synchronized neural activity, remains unknown. To address this question, our initial attempt to implement the Taylor’s law, has been hindered because neural signals at all measurement scales contain both positive and negative values, whereas Taylor’s law was restricted to non-negative ecological data [1]. To circumvent this limitation, we extend spatial Taylor’s law by replacing the sample mean with the root-mean-square (RMS) of the measurements, and apply this generalized law to detrended multivariate neural time series. The detrending operation has been performed as a necessary step to remove the trends and temporal mean from the data. Within this framework, we address our next question how synchronous or coordinated activity of the brain network influences the scaling law. We quantify synchrony using a dedicated metric of functional coordination and show analytically, and via numerical simulations using multivariate Poisson, negative binomial, uniform and gamma distributions, that the scaling exponent is inversely related to synchrony. Applying this approach to human fMRI data from three large cohorts (N = 840, ages 18–88 years) reveals distinct age-related trajectories of the scaling exponent, with healthy aging characterized by a substantial synchrony-induced reduction in this exponent. The synchrony–scaling relationship remains stable during resting-state activity but progressively shifts during naturalistic tasks across the lifespan, and is pronounced in limbic, subcortical, and cerebellar subnetworks, indicating preservation at subnetwork levels. Finally, individuals with ADHD exhibit altered synchrony–scaling coupling, highlighting the potential clinical and translational utility of this generalized scaling metric.

Spatially resolved reaction–diffusion modeling reveals effects of intracellular spatial heterogeneity on yeast galactose network dynamics

2026-10-05T14:00:00Z

by Tianyu Wu, Marie-Christin Spindler, Abner T. Apsley, Emmy Earnest, Zane R. Thornburg, Julia Mahamid, Zaida Luthey-Schulten

Eukaryotic cells are spatially organized into functionally-distinct compartments. This three-dimensional (3D) organization generates intracellular heterogeneities that can modulate regulatory dynamics. Despite this knowledge of subcellular organization, most quantitative gene-regulation models still assume a well-mixed environment in which molecules can react regardless of their spatial positions. Here, we use the well-established galactose switch in budding yeast (Saccharomyces cerevisiae) to develop spatially-resolved models that integrate experimentally-derived intracellular architectures, including chromosome organization, the endoplasmic reticulum (ER) and spatially distinct ribosome populations. We implement a hybrid stochastic–deterministic framework in which gene expression is modeled using a reaction–diffusion master equation that enforces locality (i.e., reactions occur only when molecules are in physical proximity), while metabolic and transport processes are captured by ordinary differential equations. Guided by electron microscopy and biochemical constraints, we quantify how accounting for intracellular spatial organization alters regulatory predictions in the galactose switch. We show that in present model chromosome geometry has little effect on Gal2p output, whereas ER-associated translation reduces Gal2p delivery to the plasma membrane; the largest decrease of Gal2p abundance occurs when translation of GAL2 mRNA is restricted to a population of ribosomes physically bound to the ER. Together, these results demonstrate that more realistic 3D cellular architectures and local reaction rules can qualitatively change regulatory predictions, motivating integration of intracellular organization in future whole-cell models.

Machine learning and Voronoi-based decision boundaries for Bacterial vaginosis to determine population- specific microbial interactions

2026-10-05T14:00:00Z

by Cameron G. Celeste, Carleigh C. Sokolik, Wambui Gachunga, Ivana K. Parker

In this study we utilize machine learning techniques to create predictive models and determine key bacterial interactions for the diagnosis of Bacterial vaginosis. Bacterial vaginosis (BV) is a common vaginal syndrome affecting reproductive-age women globally. It is associated with various adverse obstetric and gynecological out-comes including increased risk of sexually transmitted infections, HIV, cervical cancer, and pre-term birth. While it is known that BV is caused by a shift in abundance between Lactobacilli and anaerobic bacteria, it is unknown how gradual shifts in that balance lead towards BV status. Here we perform a rigorous comparison of machine learning architectures and feature selection methods used to train models on 16s rRNA data of patients presenting with BV. Using the highest-performing models, we employ explainable AI methods to determine the most important bacteria for BV diagnosis. Furthermore, we implement Voronoi-based decision boundaries to show how the relative abundances between pairs of these bacteria results in BV positive or BV negative outcomes. Results: We find that support vector machine and random forest models in combination with feature selection predict BV diagnosis with the most balanced accuracy. Using those models, we identify four Lactobacilli spp and six anaerobes to be key in to be key to the diagnosis of BV. The determination of key bacteria can inform BV diagnostics and pathogenesis research to species that have previously eluded scientific focus. Additionally, decision boundary plots offer a diagnostic point of reference for how the relative abundances of key vaginal flora are indicative of BV outcomes.

A graph-attentive GAN for rare-cell-aware single-cell RNA-seq data generation

2026-10-05T14:00:00Z

by Ritwik Ganguly, Sana Aafrine, Sk Md Mosaddek Hossain, Sumanta Ray

A central challenge in downstream single-cell RNA sequencing (scRNA-seq) analysis is the high-dimensional, small-sample (HDSS) regime, often compounded by class imbalance from rare cell types. These factors hinder robust feature (gene) selection and cell clustering and limit the realism of samples generated by existing simulators. We introduce GARAGE, a Graph-Attentive RAre-cell aware single-cell data GEneration that augments the generator’s input with a small, attention-weighted ‘leakage’ of real cells in addition to prior noise. Specifically, we build a k-nearest-neighbour cell graph and use a graph attention network (GAT) to prioritize nodes that likely represent under-sampled (rare) subpopulations; these high-attention cell embeddings are injected into the generator input to steer synthesis toward biologically plausible regions of the data manifold while respecting cell-type proportions. This attention-guided leakage accelerates training, reduces mode dropping, and yields realistic synthetic cells that preserve rare-cell structure. Across real scRNA-seq benchmarks, GARAGE improves downstream feature selection and clustering compared with state-of-the-art baselines. In summary, GARAGE directly addresses HDSS and rarity in scRNA-seq by coupling graph attention with adversarial generation to produce high-fidelity synthetic cells that enhance downstream analyses. The corresponding software is available at: https://github.com/RitwikGanguly/GARAGE.

Robust prediction of drug combination side effects in realistic settings

2026-10-05T14:00:00Z

by Ruben Jimenez, Alberto Paccanaro

Side effects caused by drug combinations pose a major challenge in healthcare. Knowledge of these side effects is limited because often they are not detected in clinical trials, which typically involve a restricted number of participants and tested drug combinations. We introduce DCSE (Drug Combinations Side Effects), a novel machine learning method for predicting polypharmacy side effects. DCSE learns latent signatures for drugs, drug pairs, and side effects to predict the probability that a side effect occurs in a given drug combination. We first evaluate its performance in the commonly adopted experimental settings in the literature. However, these rely on balanced testing datasets and sampled negative examples, which do not capture the highly imbalanced and structured set of unknown side effects encountered in practice. Therefore, a key contribution of this paper is the introduction of more realistic experimental settings under prospective evaluations. Here, we attempt to predict side effects reported between 2009 and 2014 after training only on data available prior to that period. These evaluations include warm-start scenarios, in which some side effects are already known for a drug pair, and cold-start scenarios, in which the model predicts side effects for previously uncharacterized drug pairs. Our results indicate that DCSE consistently outperforms state-of-the-art methods, demonstrating its robustness and efficacy in real-world applications.

Effects of spatial constraints of inhibitory connectivity on the dynamical development of criticality in spiking networks

2026-10-01T14:00:00Z

by Felix Benjamin Kern, Takahisa Date, Zenas C. Chao

Neural systems are hypothesized to operate near criticality, enhancing their capacity for optimal information processing, transmission and storage capabilities. Criticality has typically been studied in spiking neural networks and related systems organized in random or full connectivity, with the balance of excitation and inhibition being a key determinant of the critical point of the system. However, given that neurons in the brain are spatially distributed, with their distances significantly influencing connectivity and signal timing, it is unclear how the spatial organization of excitatory and inhibitory connectivity influences the network’s development towards criticality. Here, we systematically constrain the distance and density of inhibitory connectivity in two-dimensional spiking networks and allow synaptic weights to develop with activity-dependent excitatory and inhibitory plasticity in the presence of a low level of stochastic intrinsic activity. We then investigate the relationship between inhibitory connectivity, synaptic weights, and the resulting network activity during and after development. We find that networks with longer-range inhibitory synapses tend towards more supercritical behavior compared to networks with a similar number of shorter-range inhibitory synapses. We show that this distance dependence is a consequence of weaker long-range synapses after development due to the presence of synaptic delays, which shift most spike pairs outside of the potentiation window of the inhibitory learning rule.

Demographic changes and behavioural responses shape vulnerability to infectious disease outbreaks

2026-10-01T14:00:00Z

by Abbie Evans, William S. Hart, Eunok Jung, Kyeongah Nah, Karla Bonic-Babic, Sung-mok Jung, Robin N. Thompson

Demographic shifts are reshaping population age structures worldwide, with implications for infectious disease dynamics. Since contact patterns, susceptibility and infectiousness often vary by age, the risk that pathogen introductions initiate a substantial outbreak depends on the population’s age distribution and associated behavioural characteristics. We develop an age-structured mathematical model to estimate the risk that a single pathogen introduction leads to sustained transmission (the probability of a major outbreak) under long-term demographic transitions, incorporating changes in age-specific contact patterns and behavioural adaptation (specifically, extended workforce participation in an ageing society). Using the Republic of Korea (projected to become the world’s oldest population by 2050) as a case study, we show that population ageing may reduce the probability of a major outbreak due to older individuals’ lower contact rates. However, this effect is attenuated for pathogens with increasing susceptibility or infectiousness with age, and by factors such as later retirement that increase contacts among older individuals. These findings demonstrate that, while outbreak risks are affected by demographic changes, they are further modified by associated behavioural responses, highlighting the importance of accounting for demographic and socio-behavioural context when assessing future infectious disease outbreak risks.

Generative diffusion models for spatiotemporal influenza forecasting

2026-09-30T14:00:00Z

by Joseph Lemaitre, Justin Lessler

Forecasting infectious disease incidence can provide important information to guide public health planning, yet it is difficult because epidemic dynamics are complex. Current mechanistic and statistical approaches often struggle to capture multimodal uncertainty or emergent trends. Influpaint adapts denoising diffusion probabilistic models to epidemic forecasting. By encoding influenza seasons as spatiotemporal images in which pixel intensity represents incidence, Influpaint learns a rich distribution of disease dynamics from a hybrid dataset of surveillance and simulated trajectories. Forecasting is formulated as a conditional generation (inpainting) task from partial observations. We show that Influpaint generates realistic, diverse epidemic trajectories and achieves forecast accuracy that is competitive with leading ensemble methods in retrospective evaluation. In real-time evaluation during the 2022–2025 U.S. CDC FluSight challenges, performance improved substantially across seasons, with highly accurate but somewhat overconfident projections in 2024–2025. The best performance was achieved with a training dataset containing 30% surveillance and 70% simulated trajectories. These results show that diffusion models can capture important spatiotemporal structure in influenza dynamics and provide a flexible framework for probabilistic infectious disease forecasting.

Modelling vasopressin synthesis and storage dynamics during prolonged osmotic challenge and recovery based on activity dependent upregulation of mRNA transcription

2026-09-30T14:00:00Z

by Duncan J. MacGregor

Hypothalamic vasopressin neurons are neuroendocrine cells which form part of the homeostatic systems that regulate plasma volume and extracellular fluid osmolarity. In response to synaptic inputs encoding osmolarity and changes in plasma volume, they generate complex and dynamically heterogeneous patterns of spikes that in turn trigger secretion of peptide hormone vasopressin from axonal terminals in the posterior pituitary. The thousands of neurons’ secretory signals generate a summed plasma vasopressin signal acting at the kidneys to regulate water loss. Vasopressin is synthesised in the neuronal cell bodies, packaged into vesicles, and transported to large stores at the pituitary terminals. Supported by activity-dependent upregulation of synthesis and transport, these stores can maintain a secretion response for several days of elevated osmolarity, experimentally tested by dehydration or salt loading. However, despite upregulated synthesis, stores gradually decline during sustained challenge, followed by a slow recovery. With no evidence of a store encoding feedback signal, previous simple modelling explained these synthesis dynamics based on activity-dependent upregulation of transcription and mRNA content. Here we integrate this idea into a detailed neuronal model, coupling spiking, secretion, and synthesis to simulate a complete neural system, from physiological input to output, and ask how a distributed neuroendocrine system can maintain hormone supply-demand balance over prolonged challenge without direct feedback from distant hormone stores. The model suggests that mRNA acts as a long-timescale memory of neuronal activity, extending temporal integration beyond spike activity and intracellular Ca²⁺ to coordinate synthesis with sustained demand.

Mixed updating in structured populations

2026-09-30T14:00:00Z

by David A. Brewster, Yichen Huang, Michael Mitzenmacher, Martin A. Nowak

Evolutionary graph theory (EGT) studies the effect of population structure on evolutionary dynamics. The vertices of the graph represent the N individuals. The edges denote interactions for competitive replacement. Two standard update rules are death-Birth (dB) and Birth-death (Bd). Under dB, an individual is chosen uniformly at random to die, and its neighbors—the individuals on adjacent vertices of the graph— compete to fill the vacancy proportional to their fitness. Under Bd, an individual is chosen for reproduction proportional to fitness, and its offspring replaces a randomly chosen neighbor on an adjacent vertex. Here we study mixed updating between those two scenarios. In each time step, with probability δ the update is dB and with the remaining probability it is Bd. We study fixation probabilities and times as functions of δ under neutral evolution and constant selection. Despite the fact that fixation probabilities and times can be increasing, decreasing, or non-monotonic in δ, we prove that nearly all unweighted undirected graphs have short fixation times and provide an efficient algorithm to estimate their fixation probabilities. Finally, we prove exact formulas for fixation probabilities on cycles, stars, and more complex structures and classify their sensitivities to δ.

Granger sensori-behavioral functional taxonomy of neuronal ensemble activity from two-photon calcium imaging data

2026-09-30T14:00:00Z

by Sahar Khosravi, Nikolas A. Francis, Patrick O. Kanold, Behtash Babadi

Understanding how neuronal populations interact to encode and transform sensory information is a fundamental challenge in computational neuroscience. Most existing studies, however, study neural encoding, behavioral readout, and functional connectivity as disjoint problems. Two-photon calcium imaging enables simultaneous recording of large neuronal ensembles in vivo, driven by diverse stimuli and eliciting distinct behaviors. However, extracting directional functional connectivity metrics as well as encoding and readout properties of neurons from such data remains difficult due to indirect and noisy observations of spiking activity, slow temporal dynamics, and the latent interplay between external stimuli and endogenous neural processes. Here, we introduce a unified conceptual and operational modeling and inference framework for directly extracting functional Granger causal (GC) effects between neurons, from external stimuli to neurons, and from neurons to behavior, from two-photon imaging data, in the sense of Granger’s temporal predictability. Inspired by the intersection information framework, we also identify neurons that encode features of sensory stimuli that inform behavioral readout. The resulting GC networks together with the taxonomy of functional sensori-behavioral relevance, which we call G-taxonomy, provides a powerful statistical analysis framework, enabled by the integration of several techniques including state-space modeling and inference, variational inference, and point processes. We applied the proposed framework to simulated and experimentally-recorded two-photon imaging from the mouse auditory cortex (A1) during both passive listening and active tone discrimination. Our simulation studies reveal significant improvement of our proposed methodology over existing techniques. Analysis of experimental data from the mouse A1 identifies distinct groups of cells with diverse sensori-behavioral relevance, as well as changes in functional connectivity associated with correct vs. incorrect behavior. In summary, this work provides a principled and data-driven methodology for uncovering directional interactions among the neurons, sensory stimuli, and behavior, all within the same statistical framework, offering new insights into how distributed cortical populations transform sensory inputs into behaviorally relevant representations.

Modeling decision dynamics disentangles working memory, cognitive control and reinforcement learning and reveals clinical differences

2026-09-30T14:00:00Z

by Krishn Bera, Alexander Fengler, Megan A. Boudewyn, Cameron S. Carter, Molly A. Erickson, James M. Gold, Steven J. Luck, J. Daniel Ragland, Andrew P. Yonelinas, Angus W. MacDonald III, Deanna M. Barch, Michael J. Frank

Various neurocognitive processes contribute to reinforcement learning (RL) and decision making, requiring careful task designs and computational modeling to disentangle them. People rely on capacity-limited working memory (WM) to rapidly and flexibly adapt behavior during learning, together with slower incremental RL processes that support robust long-term retention. The RLWM paradigm was developed to isolate these contributions by manipulating WM demands during learning. Empirical and computational advances support the separability of such processes, but only imperfectly, because a given choice can often equally be attributed to use of WM vs. RL. Here, we show that jointly modeling decision dynamics – accounting not only for choices but also response time distributions, together with hierarchical Bayesian parameter estimation – improves the ability to disentangle separable learning and cognitive control processes. Despite added complexity, models with decision dynamics improved parameter recovery and yielded accurate out-of-sample prediction for choices in a held-out test phase after learning. In contrast, models fit only to choices produced an inflation of RL learning rates and failed the out-of-sample test. Moreover, joint modeling also revealed a novel neurocognitive process by which participants proactively widen decision boundaries to increase response caution under increased WM load. Applying our model to patients with schizophrenia revealed a deficit in proactive control mechanisms as well as slowed incremental RL, which were masked by previous choice-only models, while replicating previously established WM deficits. In summary, we show how joint modeling enables accurate, model-based and mechanism-oriented computational phenotyping of psychopathologies.

Context-dependent feature modulation shapes human decision policies in approach–avoidance conflicts

2026-09-30T14:00:00Z

by Sergej A. E. Golowin, Niall W. Duncan, Faizan Shaikh, Christoph W. Korn

Sequential decisions often involve trade-offs between immediate rewards and future risks, particularly in approach–avoidance contexts. While such behavior is commonly framed in terms of optimization principles, it remains unclear how decision processes adapt across contexts. We developed a sequential foraging task in which participants made binary choices under probabilistic reward and predation risk (threat). The design dissociated reward probability from threat and enabled formal comparison of decision-relevant features (e.g., reward and threat probabilities), a multi-feature policy, and a mathematically optimal policy derived from a fully observable Markov Decision Process. Crucially, the task included two implicitly signaled conditions – approach and avoidance – that differed in how reward and threat information jointly shaped the optimal policy. In both conditions, higher reward probabilities were associated with higher predation risk. However, the balance between these competing factors differed across environments, creating contexts in which the optimal policy either favored approaching or avoiding the higher-threat option. Behavior was analyzed using hierarchical Bayesian models capturing both individual decision features and optimal action values, as well as their modulation by task context. Avoidance contexts selectively altered the weighting of decision features, with reduced reliance on reward probability. In addition, choices showed evidence for increased alignment with optimal state–action values, even after accounting for heuristic features within a shared model. This indicates that behavior more strongly reflected the value structure of the environment under avoidance, consistent with enhanced integration of decision-relevant features. These findings suggest that adaptive behavior in sequential decision-making arises from context-dependent feature reweighting alongside increased alignment with integrated value signals. This provides a parsimonious account of how humans adjust decision-making under threat, without requiring assumptions about discrete strategy shifts.

PEtab Select: Specification standard and supporting software for automated model selection

2026-09-30T14:00:00Z

by Dilan Pathirana, Frank T. Bergmann, Domagoj Doresic, Polina Lakrisenko, Sebastian Persson, Niklas Neubrand, Jens Timmer, Clemens Kreutz, Harald Binder, Marija Cvijovic, Daniel Weindl, Jan Hasenauer, with the PEtab Select community

A central question in mathematical modeling of biological systems is determining which processes are most relevant and how they can be described. There are often competing hypotheses, which yield different models. Model comparison requires parameter optimization and sampling methods. Yet, standards for the specification of model selection problems and the swift evaluation of a broad spectrum of approaches are not available. PEtab Select addresses this challenge by providing a concise, standardized specification of model selection and its associated calibration problems through a new file format standard and software package. The standard facilitates the compact representation of even very large model selection problems; in one example, billions of model alternatives. PEtab Select builds on the PEtab standard for the specification of parameter estimation problems, and enables the use of state-of-the-art modelling and calibration workflows utilizing COPASI, Data2Dynamics, PEtab.jl, and pyPESTO. PEtab Select supports common model selection criteria (e.g., Akaike and Bayesian information criteria) and can be easily extended to use others. To ensure flexibility, PEtab Select implements several model space exploration approaches, including basic brute-force, forward, and backward selection, and also advanced, flexible selection methods. PEtab Select introduces the first standardization of model selection tasks, filling a critical gap in existing computational pipelines. It constitutes an essential contribution to FAIR research software in systems biology by promoting interoperability and reusability in model selection.

GeoEPred: A multimodal structure-aware geometric deep learning framework for Gram-negative bacterial secreted effector prediction with sequence semantics

2026-09-30T14:00:00Z

by Shouzhen Song, Hua Shi, Hongfeng Wu, Dachen Liu, Yihang Lin, Nor Ashidi Mat Isa, Quan Zou, Leyi Wei

Accurate prediction of effector proteins secreted by Gram-negative bacteria is important for elucidating bacterial pathogenic mechanisms and developing precise anti-infective strategies. Although existing methods have benefited from the strong sequence feature extraction capacity of pretrained protein language models, reliance on linear sequence information alone often fails to fully capture the three-dimensional conformational signals required for virulence functions. Meanwhile, conventional structure-based methods are limited by the scarcity of experimentally resolved protein structures. To address these challenges, we propose GeoEPred, a multimodal deep learning framework designed for the synergistic modeling of protein sequence and structure to identify Gram-negative bacterial effector proteins. Specifically, the model integrates sequence-contextual embeddings from a pretrained protein language model with three-dimensional structural representations predicted by ESMFold. A feature projection network refines fine-grained sequence signals associated with effector functions, while geometric vector perceptrons characterize inter-residue orientations, distances, and local spatial topology to capture potential structural conformational motifs. To further enable effective cross-modal fusion, we design a cross-modal alignment and feature-tokenized self-attention module. This module enhances consistency between the sequence-semantic and structural-geometric spaces through contrastive learning and models associations between linear functional motifs and spatial conformational patterns at a fine-grained token level. Extensive evaluations on multiple benchmark datasets show that GeoEPred achieves better predictive performance than existing leading models in T3SE, T4SE, and T6SE prediction tasks, while maintaining stable performance in remote homolog recognition scenarios. Moreover, the modular and extensible architecture of GeoEPred demonstrates strong generalization ability and substantial application potential for genome-scale effector protein discovery.

A compartmentalized SCN model explains temperature resistance through interregional phase differences

2026-09-29T14:00:00Z

by Feng Yu, Ling Yang, Jie Yan

Temperature cycles entrain circadian rhythms in many peripheral tissues, yet the master circadian pacemaker (suprachiasmatic nucleus) in mammals remains largely insensitive to the same input. Despite their similar molecular circuitry, the mechanisms underlying the distinct temperature responses of central and peripheral clocks remain unclear. To address this, we developed a compartmentalized model of the SCN circadian clock that integrates the heat shock protein (HSP) pathway, simulating the response of the SCN network to temperature variations. Using this model, we successfully recapitulated key experimental observations, including the ability of heat stimuli to entrain peripheral clocks to an inverted phase while the SCN maintains its resistance. Further analysis reveals that intrinsic phase heterogeneity among SCN subregions underlies this entrainment resistance. Specifically, the phase difference between the dorsal and ventral SCN leads to divergent phase responses upon simultaneous stimulation, resulting in an attenuated net shift of the overall SCN phase. Moreover, the mode of thermal input, in which the dorsal and ventral regions are stimulated concurrently, may further enhance the SCN’s robustness to temperature entrainment. These results suggest that the SCN’s resistance to temperature entrainment may differ mechanistically from photic entrainment, and highlight the potential importance of its spatial structure in maintaining circadian stability.

Spatiotemporal modeling of GPCR signaling: The role of endosomal dynamics and receptor recycling

2026-09-29T14:00:00Z

by Chloé Weckel, Juliette Gourdon, Léo Darrigade, Vinesh Jugnarain, Pascale Crépieux, Eric Reiter, Frédéric Jean-Alphonse, Stefan Haar, Romain Yvinec

A growing number of G Protein-Coupled Receptors (GPCRs) have been reported to initiate signaling both at the cell surface and from intracellular compartments such as endosomes. The kinetics and spatial localization of these signals are critical determinants of cellular responses, yet receptor trafficking - including internalization, endosomal sorting, and recycling - remains a pivotal but often overlooked component of theoretical GPCR models. Here, we present a generic GPCR dynamic model that explicitly integrates receptor trafficking and signaling compartmentalization, enabling systematic characterization of the role of endosomal dynamics and receptor recycling on ligand-induced cellular responses. Then, as a case study, a model selection method with high-throughput kinetic data from the follicle-stimulating hormone receptor (FSHR) is used to reveal the impact of receptor trafficking on FSH-induced signaling responses. Although only a small fraction of FSHRs is internalized, these receptors produce a highly active endosomal response, underscoring the strong effect that any drug perturbing receptor trafficking can have. Our approach thus provides a refined tool for the pharmacological characterization of ligands and advances understanding of the spatial organization of FSHR signaling. Beyond this specific receptor, the methodology offers a generalizable strategy for modeling GPCR trafficking dynamics.

Spatial cell-cell communication inference based on cell-spot-ligand-receptor heterogeneous graphs

2026-09-28T14:00:00Z

by Xiong Li, Dongding Wu, Yuejin Zhang, Min Chen, Chong Liu

Cell-cell communication is essential for tissue homeostasis and disease-associated microenvironment remodeling. Spatial transcriptomics enables the study of communication events in situ, but single-cell-resolution inference from spot-based data remains challenging because of uncertain cell-to-spot mapping, limited use of local spatial information, and false positives from ligand-receptor co-expression. Existing methods increasingly emphasize spatial proximity in cell-cell communication inference, while integrating biologically informative signals remains important for improving the interpretability of candidate communication events. Here, we present SpaHCC, a heterogeneous graph learning framework that integrates single-cell transcriptomics, spatial transcriptomics, and prior ligand-receptor knowledge for spatial cell-cell communication inference. By jointly leveraging molecular features, spatial neighborhood information, and cellular functional states, SpaHCC prioritizes candidate communication events and supports their biological contextualization through pathway, transcription factor, and receptor-side response analyses. Benchmark and spatial consistency analyses showed that SpaHCC recovered spatially coherent candidate communication signals. In Alzheimer’s disease data, SpaHCC identified recurrent excitatory neuron-associated candidate patterns, including NRXN- and FLRT2-related modules associated with synaptic remodeling and inflammation-related transcriptional programs. Cross-patient analyses showed recurrence of selected AD-associated candidate patterns, while representative cSCC and PDAC analyses illustrated the spatial interpretability of the framework in additional disease contexts. Overall, SpaHCC provides an effective and interpretable framework for spatially resolved cell-cell communication analysis in complex diseases.

Attribution assignment for deep-generative sequence models enables interpretability analysis using positive-only data

2026-09-28T14:00:00Z

by Robert Frank, Michael Widrich, Rahmad Akbar, Günter Klambauer, Geir Kjetil Sandve, Philippe A. Robert, Victor Greiff

Generative machine learning models offer a powerful framework for therapeutic design, by efficiently exploring large spaces of biological sequences enriched for desirable properties. Unlike supervised learning methods, which require both positive and negative labeled data, generative models such as LSTMs can be trained solely on positively labeled sequences, for example, high-affinity antibodies. This is particularly advantageous in biological settings where negative data are scarce, unreliable, or biologically ill-defined. However, the lack of attribution methods for generative models has hindered the ability to extract interpretable biological insights from such models. To address this gap, we developed Generative Attribution Metric Analysis (GAMA), an attribution method for autoregressive generative models based on Integrated Gradients. We assessed GAMA using synthetic datasets with known ground truths to characterize its statistical behavior and validate its ability to recover biologically relevant features. We further demonstrated the utility of GAMA by applying it to experimental antibody-antigen binding data. GAMA enables model interpretability and the validation of generative sequence design strategies without the need for negative training data.

From observable fermentation data to hidden cell states: A modeling study of a mixotrophic Clostridium coculture under perfusion mode

2026-09-28T14:00:00Z

by Juhyeon Kim, Hangjun Cho, Jin Hong Mok, Hyeongmin Seo, Joseph Sang-Il Kwon

Microbial cocultures exhibit complex population dynamics that are difficult to interpret because internal physiological states are only partially observable. In particular, active and dormant cell states can influence system-level behavior but are rarely resolved from standard fermentation measurements. In this study, we present a hybrid modeling framework that combines structured population-state reconstruction with sparse identification of nonlinear dynamics (SINDy) to analyze a Clostridium acetobutylicum-Clostridium ljungdahlii coculture under perfusion mode. The framework estimates active Cac and Clj biomass-equivalent trajectories from observable biomass and activity measurements, reconstructs dormant populations as model-constrained latent states, and uses these states to identify extracellular metabolite dynamics. After accounting for first-principles perfusion transport, SINDy identified sparse biological reaction terms associated with organic acid turnover, solvent formation, and acetone-to-isopropanol conversion. The resulting model captured active-biomass and metabolite trajectories and suggested that the coculture dynamics are consistent with acid-associated state transitions and sequential metabolic exchange. This framework provides a transparent strategy for interpreting partially observed microbial cocultures while explicitly treating dormant biomass as a latent reconstructed state.

Using iPALM to determine protein organisation in cardiac muscle Z-discs

2026-09-28T14:00:00Z

by Oliver Umney, Alistair P. Curd, Heather L. Martin, Tarek Lewis, Anna Ah-San Tang, Thembaninkosi Gaule, Iain W. Manfield, Harikrushnan Balasubramanian, Satya Khuon, Jesse Aaron, Michelle Peckham

Sarcomeres, the basic repeating unit of striated muscle, are joined together by crosslinked actin filaments found at the boundaries of muscle sarcomeres, termed Z-discs. Z-discs play a key role in cardiac signalling and disease, however, the arrangement and function of many of the proteins present in the Z-disc remain to be understood. Here, we determined the organisation of 3 key proteins, ZASP, ɑ-actinin-2 and the Z1Z2 epitope of titin, located within the Z-disc. We fluorescently labelled these proteins in cardiac myofibrils using Adhirons specific to each protein and used interferometric photoactivated localization microscopy (iPALM) to obtain the 3D position of these proteins to a high precision (<10nm in x,y,z). We then used PERPL (Pattern Extraction from Relative Positions of Localisations: available at https://github.com/oubino/z_disk/releases/tag/v0.0.3) to analyse patterns in the relative positions of the proteins and reveal their underlying organisation. This analysis revealed that ZASP and ɑ-actinin-2 have a similar repeating organisation, but that the organisation of Z1Z2 is different.