PLoS Computational Biology最新文献

{"title":"VarPPUD: Pinpointing diagnostic variants from sets of prioritized, strong candidate variants.","authors":"Rui Yin, Alba Gutiérrez-Sacristán, Shilpa Nadimpalli Kobren, Paul Avillach","doi":"10.1371/journal.pcbi.1013414","DOIUrl":"10.1371/journal.pcbi.1013414","url":null,"abstract":"<p><p>Rare and ultra-rare genetic conditions are estimated to impact nearly 1 in 17 people worldwide, yet accurately pinpointing the diagnostic variants underlying each of these conditions remains a formidable challenge. Because comprehensive, in vivo functional assessment of all possible genetic variants is infeasible, clinicians instead consider in silico variant pathogenicity predictions to distinguish plausibly disease-causing from benign variants across the genome. However, in the most difficult undiagnosed cases, such as those accepted to the Undiagnosed Diseases Network (UDN), existing pathogenicity predictions cannot reliably discern true etiological variant(s) from other deleterious candidate variants that were prioritized through case- or family-level analyses. Pinpointing the disease-causing variant from a small pool of plausible candidates remains a largely manual effort requiring extensive clinical workups, functional and experimental assays, and eventual identification of genotype- and phenotype-matched individuals. Here, we introduce VarPPUD, a tool trained on prioritized variants from UDN cases, that leverages gene-, amino acid-, and nucleotide-level features to discern pathogenic (disease causative) variants from other damaging or deleterious variants that are unlikely to be confirmed as relevant to the disease. VarPPUD achieves a cross-validated accuracy of 79.3% and precision of 77.5% on a held-out subset of uniquely challenging UDN cases, respectively representing an average 18.6% and 23.4% improvement over nine existing state-of-the-art pathogenicity prediction tools on this task. We validate VarPPUD's ability to discriminate likely from unlikely pathogenic variants using both synthetic data generated via a GAN-based framework and a temporally held-out set of UDN patients evaluated between 2022 and 2024. The model was trained exclusively on data available through 2021 and applied without retraining to the post-2021 cohort, demonstrating strong generalizability to newly accrued cases. Finally, we show how VarPPUD can be probed to evaluate each input feature's importance and contribution toward prediction-an essential step toward understanding the distinct characteristics of newly-uncovered disease-causing variants.</p>","PeriodicalId":20241,"journal":{"name":"PLoS Computational Biology","volume":"21 9","pages":"e1013414"},"PeriodicalIF":3.6,"publicationDate":"2025-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12468739/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145126055","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Brainwide hemodynamics predict EEG neural rhythms across sleep and wakefulness in humans.","authors":"Leandro P L Jacob, Sydney M Bailes, Stephanie D Williams, Carsen Stringer, Laura D Lewis","doi":"10.1371/journal.pcbi.1013497","DOIUrl":"10.1371/journal.pcbi.1013497","url":null,"abstract":"<p><p>The brain exhibits rich oscillatory dynamics that play critical roles in vigilance and cognition, such as the neural rhythms that define sleep. These rhythms continuously fluctuate, signaling major changes in vigilance, but the widespread brain dynamics underlying these oscillations are difficult to investigate. Using simultaneous EEG and fast fMRI in humans who fell asleep inside the scanner, we developed a machine learning approach to investigate which fMRI regions and networks predict fluctuations in neural rhythms. We demonstrated that the rise and fall of alpha (8-12 Hz) and delta (1-4 Hz) power-two canonical EEG bands critically involved with cognition and vigilance-can be predicted from fMRI data in subjects that were not present in the training set. This approach also identified predictive information in individual brain regions across the cortex and subcortex. Finally, we developed an approach to identify shared and unique predictive information, and found that information about alpha rhythms was highly separable in two networks linked to arousal and visual systems. Conversely, delta rhythms were diffusely represented on a large spatial scale primarily across the cortex. These results demonstrate that EEG rhythms can be predicted from fMRI data, identify large-scale network patterns that underlie alpha and delta rhythms, and establish a novel framework for investigating multimodal brain dynamics.</p>","PeriodicalId":20241,"journal":{"name":"PLoS Computational Biology","volume":"21 9","pages":"e1013497"},"PeriodicalIF":3.6,"publicationDate":"2025-09-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12459787/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145092384","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Inheritable cell-states shape drug-persister correlations and population dynamics in cancer cells.","authors":"Anton Iyer, Adrian Alva, Adrián E Granada, Shaon Chakrabarti","doi":"10.1371/journal.pcbi.1013446","DOIUrl":"10.1371/journal.pcbi.1013446","url":null,"abstract":"<p><p>Drug tolerant persisters (DTPs) drive cancer therapy resistance by temporarily evading drug action, allowing multiple routes to eventual permanent resistance. Despite clear evidence for DTPs, the timing of their emergence, proliferative nature, and how their population dynamics arise from measured single-cell kinetics remain poorly understood. Here we use time-lapse microscopy data from two cancer cell lines, integrating single-cell and population measurements, to develop a quantitative description of drug persistence. Contrary to the expectation that increasing levels of genotoxic stress should lead to slower times to division and faster times to death, we observe minor changes in the single-cell intermitotic and death time distributions upon increasing cisplatin concentration. Yet, population decay rates increase 3-fold, suggesting a surprising independence of the overall dynamics from the measured birth and death rates. To explain this phenomenon, we argue that the observed lineage correlations and concentration-dependent decay rates imply cell-state dependent fate choices made both pre and post-cisplatin as opposed to just post-drug birth/death rate-based competitive fate choices. We demonstrate that these cell-states, present in the pre-drug ancestors of DTP and sensitive cells, exhibit no difference in cycling speed and are inherited across at least 2-3 cellular generations. Post-drug survival versus death fates are decided with high probability by these pre-existing cell-states, but get modulated to some extent by the drug, leading to a drug concentration dependent state-fate map. A stochastic model implementing these rules simultaneously recapitulates the observed decay rates and cell-fate lineage correlations. The model also demonstrates how the use of barcode diversity change before and after drug might lead to misleading interpretations of the timing of persister fate decisions. Our results provide a conceptual framework for quantifying pre versus post-drug contributions to cell fate, without requiring knowledge of the underlying molecular architecture of the heterogeneous cell states.</p>","PeriodicalId":20241,"journal":{"name":"PLoS Computational Biology","volume":"21 9","pages":"e1013446"},"PeriodicalIF":3.6,"publicationDate":"2025-09-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12469175/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145092339","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Mouse-Specific Single cell cytokine activity prediction and Estimation (MouSSE).","authors":"Azka Javaid, H Robert Frost","doi":"10.1371/journal.pcbi.1013475","DOIUrl":"10.1371/journal.pcbi.1013475","url":null,"abstract":"<p><p>The accurate cell-level characterization of cytokine activity is important for understanding the signaling processes underpinning a wide range of immune-mediated conditions such as auto-immune disease, cancer and response to infection. We previously proposed the SCAPE (Single cell transcriptomics-level Cytokine Activity Prediction and Estimation) method to address the challenges associated with cytokine activity estimation in human single cell RNA-sequencing (scRNA-seq) and spatial transcriptomics (ST) data. Here, we propose a new method MouSSE (Mouse-Specific Single cell transcriptomics level cytokine activity prediction and Estimation) for performing cytokine activity estimation in murine scRNA-seq and ST data. MouSSE estimates the cell-level activity of 86 distinct cytokines using a gene set scoring approach. The cytokine-specific gene sets used by MouSSE are constructed using experimental cytokine stimulation data from the Immune Dictionary and cell-level scores are computed using a modification of the Variance-adjusted Mahalanobis (VAM) technique that supports both positive and negative gene weights. MouSSE is validated using data from both the Immune Dictionary via stratified cross-validation and external scRNA-seq and ST datasets against 10 cytokine activity estimation methods. These results demonstrate that MouSSE outperforms comparable methods for cell-level cytokine activity estimation in mouse scRNA-seq and ST data. An example vignette and installation instructions for the MouSSE R package are provided at https://github.com/azkajavaid/MousseR-package.</p>","PeriodicalId":20241,"journal":{"name":"PLoS Computational Biology","volume":"21 9","pages":"e1013475"},"PeriodicalIF":3.6,"publicationDate":"2025-09-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12469201/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145092362","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Modeling neuron-astrocyte interactions in neural networks using distributed simulation.","authors":"Han-Jia Jiang, Jugoslava Aćimović, Tiina Manninen, Iiro Ahokainen, Jonas Stapmanns, Mikko Lehtimäki, Markus Diesmann, Sacha J van Albada, Hans Ekkehard Plesser, Marja-Leena Linne","doi":"10.1371/journal.pcbi.1013503","DOIUrl":"10.1371/journal.pcbi.1013503","url":null,"abstract":"<p><p>Astrocytes engage in local interactions with neurons, synapses, other glial cell types, and the vasculature through intricate cellular and molecular processes, playing an important role in brain information processing, plasticity, cognition, and behavior. This study advances understanding of local interactions and self-organization of neuron-astrocyte networks and contributes to the broader investigation of their potential relationship with global activity regimes and overall brain function. We present six new contributions: (1) the development of a new model-building framework for neuron-astrocyte networks, (2) the introduction of connectivity concepts for tripartite neuron-astrocyte interactions in biological neural networks, (3) the design of a scalable architecture capable of simulating networks with up to a million cells, (4) a formalized description of neuron-astrocyte modeling that facilitates reproducibility, (5) the integration of experimental data to a greater extent than existing studies, and (6) simulation results demonstrating how neuron-astrocyte interactions drive the emergence of synchronization in local neuronal groups. Specifically, we develop a new technology for representing astrocytes and their interactions with neurons in distributed simulation code for large-scale spiking neuronal networks. This includes an astrocyte model with calcium dynamics, an extended neuron model receiving calcium-dependent signals from astrocytes, and a parallelized connectivity generation scheme for tripartite interactions between pre- and postsynaptic neurons and astrocytes. We verify the efficiency of our reference implementation through benchmarks varying in computing resources and network sizes. Our in silico experiments reproduce experimental data on astrocytic effects on neuronal synchronization, demonstrating that astrocytes consistently induce local synchronization in groups of neurons across various connectivity schemes and global activity regimes. By adjusting the strength of neuron-astrocyte interactions, we can switch the global activity regime from asynchronous to network-wide synchronization. This work represents an advancement in neuron-astrocyte modeling, introducing a novel framework that enables large-scale simulations of astrocytic influence on neuronal networks.</p>","PeriodicalId":20241,"journal":{"name":"PLoS Computational Biology","volume":"21 9","pages":"e1013503"},"PeriodicalIF":3.6,"publicationDate":"2025-09-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12494295/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145092321","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Optimizing crop varietal mixtures for viral disease management: A case study on cassava virus epidemics.","authors":"Israël Tankam Chedjou, Ruairí Donnelly, Christopher A Gilligan","doi":"10.1371/journal.pcbi.1012842","DOIUrl":"10.1371/journal.pcbi.1012842","url":null,"abstract":"<p><p>Cassava viral diseases, including Cassava Mosaic Disease (CMD) and Cassava Brown Streak Disease (CBSD), pose significant threats to global food security, particularly in sub-Saharan Africa. This study explores the potential of varietal mixtures as a sustainable disease management strategy by introducing CropMix, a novel web-based application. The application encodes a flexible insect-borne plant pathogen transmission model to predict and optimize yields under scenarios of varietal mixtures. For instance, we use the application to evaluate the ability of virus-resistant cassava varieties (which may have lower yields in the absence of disease) to protect more susceptible varieties against CMD and CBSD, and we also consider mixtures involving tolerant varieties and non-host crops. For CMD, the high transmission rates of cassava mosaic virus (genus Begomovirus) limits the efficacy of mixtures, with susceptible monocultures emerging as more productive than susceptible-resistant mixtures whatever the whitefly pressure. In contrast, for CBSD, varietal mixtures demonstrate substantial benefits, with resistant varieties shielding susceptible ones and mitigating severe yield losses under moderate or high insect pressure. Management strategies involving non-host crops and complementary control measures, such as roguing, can further enhance outcomes. The model's simplicity and adaptability make it suitable for tailoring recommendations to diverse insect-borne crop viral diseases and agroecological contexts. The study emphasizes the need for integrating real-world data and participatory frameworks to refine and implement disease management strategies. We discuss the critical balance between agronomic potential and farmer acceptability, underscoring the importance of collaborative efforts to ensure sustainable cassava production.</p>","PeriodicalId":20241,"journal":{"name":"PLoS Computational Biology","volume":"21 9","pages":"e1012842"},"PeriodicalIF":3.6,"publicationDate":"2025-09-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12469245/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145086891","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"wavess: An R package for simulation of adaptive within-host virus sequence evolution.","authors":"Narmada Sambaturu, Zena Lapp, Fernando D K Tria, Ethan Romero-Severson, Carmen Molina-París, Thomas Leitner","doi":"10.1371/journal.pcbi.1013437","DOIUrl":"10.1371/journal.pcbi.1013437","url":null,"abstract":"<p><p>Simulating within-host virus sequence evolution allows for the investigation of factors such as the role of recombination in virus diversification and the impact of selective pressures on virus evolution. Here, we provide a new software to simulate virus within-host evolution called wavess (within-host adaptive virus evolution sequence simulator), a discrete-time individual-based model and a corresponding user-friendly R package. The underlying model simulates recombination, a latent infected cell reservoir, and three forms of selection: conserved sites fitness and replicative fitness in comparison to a reference sequence, and immune fitness including cross-reactivity imposed by a co-evolving immune response. In the R package, we also provide functions to generate model inputs from empirical data, as well as functions to analyze the simulation outputs. At user-defined time points, the software returns various counts related to the virus population(s) and a set of sampled virus sequences. We applied this model to investigate the selection pressures on HIV-1 env sequences longitudinally collected from 11 individuals. The best-fitting immune cost differed across individuals, mirroring the real-world expectation of heterogeneous immune responses among human hosts. Furthermore, the phylogenies reconstructed from these simulated sequences were similar to the phylogenies reconstructed from the real sequences for all summary statistics tested. To our knowledge, compared to other similar models, wavess has been more rigorously validated against real within-host virus sequences, and is the first to be implemented as an R package. The wavess R package can be downloaded from https://github.com/MolEvolEpid/wavess.</p>","PeriodicalId":20241,"journal":{"name":"PLoS Computational Biology","volume":"21 9","pages":"e1013437"},"PeriodicalIF":3.6,"publicationDate":"2025-09-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12459828/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145086939","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Opportunities for machine learning to predict cross-neutralization in FMDV serotype O.","authors":"Dennis N Makau, Jonathan Arzt, Kimberly VanderWaal","doi":"10.1371/journal.pcbi.1013491","DOIUrl":"10.1371/journal.pcbi.1013491","url":null,"abstract":"<p><p>Accurately estimating cross-neutralization between serotype O foot-and-mouth disease viruses (FMDVs) is critical for guiding vaccine selection and disease management. In this study, we developed a machine learning approach to estimate r1 values-an established measure of antigenic similarity-using VP1 sequence data and published virus neutralization titer (VNT) results. Our dataset comprised 108 serum-virus pairs representing 73 distinct FMDV strains. We applied Boruta feature selection and random forest classifiers, optimizing model performance through tenfold cross-validation and sub-sampling to address class imbalance. Predictors included pairwise amino acid distances, site-specific polymorphisms, and differences in potential N-glycosylation sites. Using a 0.3 r1 threshold to define cross-neutralization, the final model achieved high accuracy (0.96), sensitivity (0.93), and specificity (0.96) in training, and performed robustly on independent test sets - accuracy was 0.75 (95% CI 0.60 and 0.90), F1 score 0.86% and PPV 0.77. Importantly, key VP1 residues-positions 48, 100, 135, 150, and 151-emerged as strong predictors of antigenic relationships. Our results demonstrate the utility of integrating routinely generated genomic data with machine learning to inform vaccine candidate selection and anticipate immune interactions among circulating FMDV strains. This approach offers a practical tool for accelerating vaccine decision-making and can be adapted to other FMDV serotypes. The latest version of the r1 predictive model is available for access via a Shiny dashboard (https://dmakau.shinyapps.io/PredImmune-FMD/).</p>","PeriodicalId":20241,"journal":{"name":"PLoS Computational Biology","volume":"21 9","pages":"e1013491"},"PeriodicalIF":3.6,"publicationDate":"2025-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12456779/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145081538","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Ligands binding diffusively to protein target act as inhibitors of protein-protein interactions.","authors":"William Jeffries, Bryan M Delfing, Xavier E Laracuente, Xingyu Luo, Audrey Olson, Kenneth W Foreman, Kyung Hyeon Lee, Greg Petruncio, Vito De Benedictis, Mikell Paige, Kylene Kehn-Hall, Christopher Lockhart, Dmitri K Klimov","doi":"10.1371/journal.pcbi.1013495","DOIUrl":"10.1371/journal.pcbi.1013495","url":null,"abstract":"<p><p>Nuclear localization signal (NLS) sequence from capsid protein of Venezuelan equine encephalitis virus (VEEV) binds to importin-α transport protein and clogs nuclear import. Prevention of viral NLS binding to importin-α may represent a viable therapeutic route. Here, we investigate the molecular mechanism by which two diffusively binding inhibitors, DP9 and DP9o, interfere with the binding of VEEV's NLS peptide to importin-α. Our study uses all-atom replica exchange molecular dynamics simulations, which probe the competitive binding of the VEEV NLS fragment, the coreNLS peptide, and the inhibitors to importin-α. Our previous simulations of non-competitive binding of the coreNLS, in which it natively binds to importin-α, are used as a reference. Both inhibitors abrogate native peptide binding and reduce the fraction of its native interactions, but they fail to prevent its non-native binding to importin-α. As a result, these inhibitors turn the coreNLS into diffusive binder, which adopts a manifold of non-native binding poses. Competition from the inhibitors compromises the free energy of coreNLS binding to importin-α showing that they reduce its binding affinity. The inhibition mechanism is based on masking the native binding interactions formed by the coreNLS amino acids. Surprisingly, ligand interference with the binding interactions formed by importin-α amino acids contributes little to inhibition. We show that DP9 is a stronger inhibitor than DP9o. By comparative analysis of DP9 and DP9o interactions we determine the atomistic reason for a relative \"success\" of DP9, which is due to the intercalation of this inhibitor between the side chains of NLS lysine residues. To test our simulations, we performed AlphaScreen experiments measuring IC50 values for the inhibitors. AlphaScreen data confirmed in silico ranking of the inhibitors. By combining our recent studies, we discuss the putative mechanism by which diffusively binding inhibitors impact protein-protein interactions.</p>","PeriodicalId":20241,"journal":{"name":"PLoS Computational Biology","volume":"21 9","pages":"e1013495"},"PeriodicalIF":3.6,"publicationDate":"2025-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12456801/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145081470","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Reconstructing noisy gene regulation dynamics using extrinsic-noise-driven neural stochastic differential equations.","authors":"Jiancheng Zhang, Xiangting Li, Xiaolu Guo, Zhaoyi You, Lucas Böttcher, Alex Mogilner, Alexander Hoffmann, Tom Chou, Mingtao Xia","doi":"10.1371/journal.pcbi.1013462","DOIUrl":"10.1371/journal.pcbi.1013462","url":null,"abstract":"<p><p>Proper regulation of cell signaling and gene expression is crucial for maintaining cellular function, development, and adaptation to environmental changes. Reaction dynamics in cell populations is often noisy because of (i) inherent stochasticity of intracellular biochemical reactions (\"intrinsic noise\") and (ii) heterogeneity of cellular states across different cells that are influenced by external factors (\"extrinsic noise\"). In this work, we introduce an extrinsic-noise-driven neural stochastic differential equation (END-nSDE) framework that utilizes the Wasserstein distance to accurately reconstruct SDEs from stochastic trajectories measured across a heterogeneous population of cells (extrinsic noise). We demonstrate the effectiveness of our approach using both simulated and experimental data from three different systems in cell biology: (i) circadian rhythms, (ii) RPA-DNA binding dynamics, and (iii) NF[Formula: see text]B signaling processes. Our END-nSDE reconstruction method can model how cellular heterogeneity (extrinsic noise) modulates reaction dynamics in the presence of intrinsic noise. It also outperforms existing time-series analysis methods such as recurrent neural networks (RNNs) and long short-term memory networks (LSTMs). By inferring cellular heterogeneities from data, our END-nSDE reconstruction method can reproduce noisy dynamics observed in experiments. In summary, the reconstruction method we propose offers a useful surrogate modeling approach for complex biophysical processes, where high-fidelity mechanistic models may be impractical.</p>","PeriodicalId":20241,"journal":{"name":"PLoS Computational Biology","volume":"21 9","pages":"e1013462"},"PeriodicalIF":3.6,"publicationDate":"2025-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12513633/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145081464","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}