NPJ Systems Biology and Applications最新文献

Big biological datasets, such as gene expression profiles, often contain redundant features that degrade model performance and limit generalization across independent datasets with complexities like class imbalance and hidden sub-clusters. To overcome challenges, we present 'FLASH', a novel feature selection method combining filtration and heuristic-based systematic elimination. FLASH generates random samples and computes p-values for each feature using multiple statistical tests (t-test, ANOVA, Wilcoxon Rank-Sum, Brunner-Munzel, Mann-Whitney). Features are scored by aggregating significant p-values across samples. The coefficient from the machine learning model with the highest accuracy on the filtered features is used to rank them. Recursive elimination with cross-validation systematically removes features while monitoring accuracy. The final subset is selected based on the highest performance during elimination, to achieve effective feature selection. We show that our method preserves predictive performance on independent datasets. Our comprehensive evaluation across diverse datasets showed that FLASH outperforms the compared feature selection methods dRFE, Mutual information, MRMR, ElasticNet, NeuralNet, Permutation test and SAGA within the scope of our tested datasets and evaluation settings. Additionally, features selected by FLASH demonstrated greater biological relevance, as evidenced by higher overlap with disease-associated genes from DisGeNET in an independent dataset.

Escalated doses of radiotherapy associate with improved local control and overall survival (OS) in intrahepatic cholangiocarcinoma (iCCA), but personalization remains limited because conventional size-based CT criteria correlate poorly with outcomes. We hypothesized that quantitative enhancement measurements would better predict clinical outcomes and guide individualized RT optimization. In a retrospective cohort of 154 patients, we analyzed pre- and post-RT CT scans using quantitative European Association for Study of Liver (qEASL) to derive viable tumor volumes, comparing enhancement-based metrics with size-based RECIST and linking them to outcomes via survival and mathematical modeling. Change in enhancement volume was strongly associated with OS after adjustment, outperforming RECIST, and a ≥ 33% reduction optimally distinguished responders. From modeling analyses, the patient-specific tumor growth rate parameter emerged as the dominant mechanistic predictor, achieving 80.5% classification accuracy. Importantly, CT-derived mathematical parameters from this framework may inform RT planning and dose adaptation, particularly for resistant tumors, by bridging imaging with mechanistic insight.

Drug repurposing (DR) has gained significant attention as a cost-effective strategy for identifying new therapeutic uses for existing drugs. Heterogeneous network-based methods are particularly promising because they exploit complex biological interactions. However, comprehensive benchmarking across multiple datasets is still needed to assess their reliability and generalizability. We systematically evaluate ten advanced heterogeneous network-based DR methods across eight datasets, including six publicly available and two newly introduced drug-disease datasets. The methods include (i) matrix factorization: NMF, NMF-PDR, NMF-DR, VDA-GKSBMF, (ii) matrix completion: BNNR, OMC, HGIMC, (iii) recommendation systems: IBCF, LIBMF, and (iv) a deep learning approach: DRDM. Performance is assessed using the area under the receiver operating characteristic (AUC) and precision-recall curve (AUPR). We also analyze the impact of data sparsity and compare findings with previous benchmarking studies. Our results reveal that OMC consistently achieves the highest AUC and AUPR across most datasets. BNNR, DRDM, HGIMC, VDA-GKSBMF, and NMF-PDR, also demonstrate competitive performance, with NMF-PDR outperforming other NMF-based approaches. We find that differences in cross-validation strategies substantially impact reported AUPR values, with previous studies overestimating performance by omitting many negative instances. This work provides a reliable benchmarking framework and new datasets, offering insights for future research in DR.

Stroke disrupts brain function beyond focal lesions, altering multiscale temporal dynamics essential for information processing. We investigated intrinsic neural timescales (INT) and other properties of long-range temporal correlations, using longitudinal fMRI data from 15 ischemic stroke patients across 6 months, and compared them to age-matched controls. Results show that stroke patients exhibited significantly prolonged INT in multiple cortical regions, reflecting slowed temporal dynamics and disrupted hierarchy. These dynamic changes persisted through recovery and were more pronounced in patients with poor outcomes, especially within cognitive control networks. Computational modeling suggested that stroke-induced INT prolongation driven by heightened neuronal excitability reflects a dynamic shift towards criticality. Our findings position long-range temporal correlations and INT as potential biomarkers for monitoring and predicting functional recovery. This framework provides a novel perspective on stroke-induced brain changes and suggests avenues for targeted neurorehabilitation using interventions aiming at restoring intrinsic temporal dynamics.

Cellular processes evolve dynamically across time and space. Single-cell and spatial omics technologies have provided high-resolution snapshots of gene expression, greatly expanding the capability to characterize cellular states. This review summarizes recent modeling strategies for time-series and spatiotemporal transcriptomic data, emphasizing links between dynamical systems, generative modeling, and biological insight. These approaches illustrate how computational tools can deepen our understanding of the dynamic nature of single cells.

This paper reviews the design and application of mammalian synthetic gene circuits for biopharmaceutical manufacturing. It discusses key design principles and outlines transcription factors, DNA-binding proteins, and RNA as input and regulatory modules, while also presenting computational modelling as a driver for circuit optimisation. The review highlights potential applications towards the production of next-generation biotherapeutics by providing examples on monoclonal antibody glycosylation control, CAR-T cell therapy safety, and gene therapy viral vector yields.

Circadian clocks orchestrate behavior, physiology, and metabolism in harmony with the Earth's 24-h cycle. Low temperatures are known to disrupt circadian clocks in plants and poikilotherms; however, their effects on human circadian rhythms remain poorly understood. Here, we demonstrate that cold exposure abolishes the circadian rhythm in cultured human cells through diminishing the oscillation amplitude, which was restored upon rewarming. In addition, the oscillation amplitude of the 24-h temperature cycles was enhanced through resonance, reflecting the intrinsic frequency of the circadian clock. From a theoretical perspective, these dynamics correspond to Hopf bifurcation, which is confirmed by a mathematical model for the mammalian circadian clock. In contrast, the circadian amplitude of human hair follicle cells was not significantly sensitive to temperature changes. These observations suggest a potential evolutionary advantage of maintaining Hopf bifurcation despite robust homeostasis.

Scientific literature is being published at an exponential rate, including in the field of mammalian cell bioprocessing. At the same time, the research landscape is becoming more diverse, with the emergence of multiple specialised subfields. This rise in information availability as well as broadening of research fields has a direct impact on ease of information retrieval. While this growth offers valuable insights, it also makes information retrieval more complex. Developing effective literature search queries has become increasingly challenging. This work discusses the process of literature query search refinement and the nuances of maintaining search sensitivity and specificity in the context of multi-omics research for next-generation mammalian cell bioprocessing.

Ataxia-Telangiectasia (A-T) is a rare genetic disorder caused by ATM mutations, leading to impaired DNA repair, oxidative stress, and neurodegeneration. We developed a computational model of ATM-mediated signaling using ordinary differential equations in COPASI, capturing key processes including DNA damage sensing, cell cycle regulation, autophagy, and oxidative stress response. The model simulates physiological, ATM-deficient, and drug-treated conditions to explore repurposing strategies. We evaluated the effects of spermidine, omaveloxolone, and HDAC4 inhibition, revealing mechanisms by which these compounds modulate dysfunctional signaling. Sensitivity and stability analyses confirmed the model's robustness, while enrichment analysis validated involvement of key pathways. Our results highlight the synergistic potential of combining autophagy activation and epigenetic modulation to partially restore homeostasis in ATM-deficient cells. This work introduces a generalizable modeling framework for simulating disease-specific signaling dysfunction and identifying therapeutic interventions, illustrating the value of computational systems biology in rare disease drug repurposing.

The adaptive response of cancer cells to hypoxia, a key microenvironmental factor in solid tumors, is orchestrated by Hypoxia-inducible factor 1 (HIF-1). Recent evidence indicate that oxygen tension in tumor is dynamic, with hypoxia being frequently unstable, or cycling. Cycling hypoxia is associated with specific phenotypic outcomes for the cancers. Transcriptomic analysis shows that for most genes, expression changes in cycling hypoxia lie expectedly in between the change caused by stable hypoxia, suggesting multi-cycle averaging of dosage in the oxygen tension, and likely HIF-1 induced transcription. However, a small subset of genes show an oscillation/cycling hypoxia specific response, suggesting that the transcriptional machinery of these genes may interpret cycling HIF-1 activity differently from stably high HIF-1 activity. Here, we model a gene regulatory circuit, the incoherent feed-forward loops (IFFLs) to illustrate that there are parameter regimes in such genetic circuits where oscillatory specific transcription is plausible. In these IFFL models, HIF-1 regulates gene transcription of a target gene directly, as well indirectly via another transcription factor with an opposite effect on gene transcription. This IFFL circuit is able to generate gene expression of certain target genes that is more extreme than either normoxia or stable hypoxia, and this nonlinear IFFL behavior can result from either the dynamic nature or even the intermediate, time averaged hypoxic signal Supplementary Information 1 (Steady state analysis of IFFL circuits). This gene circuit also allows us to search for plausible signaling intermediaries involved in the IFFL mediated cycling hypoxic response. Finally, we present experimental evidence suggesting that HIF-1 can form IFFLs with two key transcription factors p53, and Notch1, resulting in cycling hypoxia-specific gene expression linked to breast cancer progression and poor prognosis. Our work aims to draw attention to genetic circuits as plausible mechanisms where temporal fluctuations in the tumor microenvironment may directly inform downstream transcription. These ideas could identify hitherto unknown HIF-1 driven mechanism of cancer progression contributing to emergent tumor heterogeneity.