PLoS Computational Biology最新文献

{"title":"Spatial Transcriptomics As Rasterized Image Tensors (STARIT) characterizes cell states with subcellular molecular heterogeneity.","authors":"Dee Velazquez, Caleb Hallinan, Roujin An, Kalen Clifton, Jean Fan","doi":"10.1371/journal.pcbi.1014845","DOIUrl":"10.1371/journal.pcbi.1014845","url":null,"abstract":"<p><p>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.</p>","PeriodicalId":20241,"journal":{"name":"PLoS Computational Biology","volume":"22 10","pages":"e1014845"},"PeriodicalIF":3.6,"publicationDate":"2026-10-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13657327/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149114180","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":"Quantifying antibiotic susceptibility and inoculum effects using transient dynamics of Pseudomonas aeruginosa.","authors":"Sarah Sundius, Jennifer Farrell, Kelly L Eick, Rachel Kuske, Sam P Brown","doi":"10.1371/journal.pcbi.1014827","DOIUrl":"10.1371/journal.pcbi.1014827","url":null,"abstract":"<p><p>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.</p>","PeriodicalId":20241,"journal":{"name":"PLoS Computational Biology","volume":"22 10","pages":"e1014827"},"PeriodicalIF":3.6,"publicationDate":"2026-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13657402/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149114134","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":"A graph-attentive GAN for rare-cell-aware single-cell RNA-seq data generation.","authors":"Ritwik Ganguly, Sana Aafrine, Sk Md Mosaddek Hossain, Sumanta Ray","doi":"10.1371/journal.pcbi.1014601","DOIUrl":"https://doi.org/10.1371/journal.pcbi.1014601","url":null,"abstract":"<p><p>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.</p>","PeriodicalId":20241,"journal":{"name":"PLoS Computational Biology","volume":"22 10","pages":"e1014601"},"PeriodicalIF":3.6,"publicationDate":"2026-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13638049/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149114095","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":"Machine learning and Voronoi-based decision boundaries for Bacterial vaginosis to determine population- specific microbial interactions.","authors":"Cameron G Celeste, Carleigh C Sokolik, Wambui Gachunga, Ivana K Parker","doi":"10.1371/journal.pcbi.1014767","DOIUrl":"https://doi.org/10.1371/journal.pcbi.1014767","url":null,"abstract":"<p><p>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.</p>","PeriodicalId":20241,"journal":{"name":"PLoS Computational Biology","volume":"22 10","pages":"e1014767"},"PeriodicalIF":3.6,"publicationDate":"2026-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149114157","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Spatially resolved reaction-diffusion modeling reveals effects of intracellular spatial heterogeneity on yeast galactose network dynamics.","authors":"Tianyu Wu, Marie-Christin Spindler, Abner T Apsley, Emmy Earnest, Zane R Thornburg, Julia Mahamid, Zaida Luthey-Schulten","doi":"10.1371/journal.pcbi.1014811","DOIUrl":"https://doi.org/10.1371/journal.pcbi.1014811","url":null,"abstract":"<p><p>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.</p>","PeriodicalId":20241,"journal":{"name":"PLoS Computational Biology","volume":"22 10","pages":"e1014811"},"PeriodicalIF":3.6,"publicationDate":"2026-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149114173","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"DeCTCF: Decoding CTCF binding sequences by leveraging predicted epigenomic features.","authors":"Lu Chai, Jie Gao, Tinghe Guo, Teer Ba, Zihan Li, Junjie Liu, Yong Wang, Lirong Zhang","doi":"10.1371/journal.pcbi.1014848","DOIUrl":"10.1371/journal.pcbi.1014848","url":null,"abstract":"<p><p>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.</p>","PeriodicalId":20241,"journal":{"name":"PLoS Computational Biology","volume":"22 10","pages":"e1014848"},"PeriodicalIF":3.6,"publicationDate":"2026-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13652320/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149114056","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":"Bioactivity-driven prediction of antibacterial synergy using machine learning models.","authors":"Hannie Yousefabadi, Mahya Mehrmohamadi","doi":"10.1371/journal.pcbi.1014849","DOIUrl":"10.1371/journal.pcbi.1014849","url":null,"abstract":"<p><strong>Motivation: </strong>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.</p><p><strong>Results: </strong>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.</p>","PeriodicalId":20241,"journal":{"name":"PLoS Computational Biology","volume":"22 10","pages":"e1014849"},"PeriodicalIF":3.6,"publicationDate":"2026-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13649120/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149114057","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":"A generalized scaling law reveals synchrony-driven reorganization of brain dynamics across human lifespan.","authors":"Suman Saha, Chittaranjan Hens, Priyanka Chakraborty, Tomasz Kapitaniak, Gustavo Deco, Arpan Banerjee, Syamal Kumar Dana, Dipanjan Roy","doi":"10.1371/journal.pcbi.1014821","DOIUrl":"10.1371/journal.pcbi.1014821","url":null,"abstract":"<p><p>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. 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.</p>","PeriodicalId":20241,"journal":{"name":"PLoS Computational Biology","volume":"22 10","pages":"e1014821"},"PeriodicalIF":3.6,"publicationDate":"2026-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13652345/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149114093","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":"Robust prediction of drug combination side effects in realistic settings.","authors":"Ruben Jimenez, Alberto Paccanaro","doi":"10.1371/journal.pcbi.1013619","DOIUrl":"10.1371/journal.pcbi.1013619","url":null,"abstract":"<p><p>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.</p>","PeriodicalId":20241,"journal":{"name":"PLoS Computational Biology","volume":"22 10","pages":"e1013619"},"PeriodicalIF":3.6,"publicationDate":"2026-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13657403/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149114175","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":"Effects of spatial constraints of inhibitory connectivity on the dynamical development of criticality in spiking networks.","authors":"Felix Benjamin Kern, Takahisa Date, Zenas C Chao","doi":"10.1371/journal.pcbi.1014798","DOIUrl":"10.1371/journal.pcbi.1014798","url":null,"abstract":"<p><p>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.</p>","PeriodicalId":20241,"journal":{"name":"PLoS Computational Biology","volume":"22 10","pages":"e1014798"},"PeriodicalIF":3.6,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13645236/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149114129","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}