NPJ Systems Biology and Applications最新文献

Acute myeloid leukemia (AML) is a clinically aggressive hematologic malignancy driven by complex genetic and epigenetic aberrations. Circular RNAs (circRNAs), characterized by covalently closed structures and exceptional stability, have emerged as promising diagnostic biomarkers. However, existing circRNA-based predictive models largely depend on differential expression, overlooking the potential impact of higher-order chromatin organization on circRNA formation and function. Here, we propose a machine learning framework that integrates three-dimensional (3D) genome architecture to refine circRNA selection for AML prediction. By mapping 9,565 circRNAs onto a 3D chromatin model reconstructed from Hi-C data, we analyzed their spatial clustering and biological pathway enrichment. Eighteen pathways exhibited significant 3D aggregation of circRNAs, enabling radial stratification based on nuclear localization. Five circRNA panels were designed using complementary strategies combining expression, pathway, and spatial features. Cross-validation and external validation across six machine learning algorithms showed that the panel derived from the fifth radial layer (Panel-3DG-Radius5) achieved the most robust and consistent performance (ROC-AUC > 0.99). Integrating 3D genomic context reduced feature collinearity while enhancing biological interpretability. Overall, our study establishes a 3D genome-informed paradigm for circRNA biomarker discovery, demonstrating that spatial genome organization can substantially improve the precision and robustness of AML predictive modeling.

Tumors with high copy number alteration (CNA) burden often respond poorly to immune checkpoint inhibitor therapy. However, how CNAs affect the anti-cancer immune response remains unclear. To address this, we set out to capture the transcriptional effects of CNAs and define a comprehensive landscape of immune-related transcriptional patterns. Hereto, we applied consensus independent component analysis to 294,159 bulk transcriptomic profiles. We demonstrated the predictive power of these patterns for immunotherapy response, their reproducibility across platforms, and their applicability to bulk, single-cell, and spatial transcriptomic data. Our analysis identified both novel inverse and positive associations between high CNA burden and immune-related transcriptional patterns across various cancer types. For example, higher CNA burden correlated with increased immunosuppression, including IL-17-producing cells and regulatory T cells. This resource, along with the classification of these transcriptional patterns as immune-suppressive and immune-stimulatory, may provide insights to improve immunotherapy efficacy in tumors with high CNA burden.

Optic glioma, a slow-growing tumor, is associated with Neurofibromatosis type 1 (NF1) mutations and increased midkine (MDK) production. A connection between asthma and optic glioma has previously been observed, but the mechanisms are unclear. To elucidate the role of asthma in the regulation of glioma formation, we investigated the role of T cells and the subsequent pathways in the regulation of microglia, a key player in the glioma tumor microenvironment (TME). While asthma is often linked to chronic inflammation, our mathematical analysis and experimental evidence suggest that inflammation can play a key role in suppressing the proliferation of optic glioma cells via immune reprogramming of T cells and the delicate control of signaling networks in microglia. Our mathematical model unveils the complex interactions between tumor and immune cells in optic glioma. Our results indicate that asthma-induced T cell reprogramming inhibits tumor growth by promoting the release of decorin and a subsequent suppression of CCR8 and the intercellular binding kinetics in microglia, followed by blocking of CCL5 production in TME via suppression of NFκB. We developed anti-cancer strategies by leveraging this asthma-induced immune regulation.

How do cancer cells respond to their environment, and what are the key regulators behind their behaviour? While immune cell reprogramming in the tumour microenvironment (TME) has been extensively studied, the dynamic regulatory changes within cancer cells in response to interactions with immune cells remain poorly understood. In Chronic Lymphocytic Leukaemia (CLL), this knowledge gap limits our ability to fully grasp the disease progression and to design effective, personalised interventions. To tackle this, we combine time-series transcriptomics with data-driven gene regulatory network (GRN) inference to uncover the temporal regulatory mechanisms driving CLL cell behaviour within a reconstituted in vitro TME. Using cultures of peripheral blood from CLL patients or of purified patient-derived CLL cells, we profile gene expression across five time points spanning 14 days under these experimental conditions. By inferring GRNs from transcription factor activity, we capture patient-specific and temporally resolved regulatory interactions that highlight how immune signals drive cancer cell phenotypic changes. Our network analysis reveals distinct gene modules associated with critical processes such as cytokine signalling, metabolic reprogramming and differentiation, hallmarks of immune-cancer cell interaction. Intriguingly, we found that while the presence of immune cells in the environment significantly alters CLL cell activation, their survival trajectories are predominantly governed by intrinsic features. This study not only offers mechanistic insights into how immune cell presence influences CLL cell fate but also presents a robust computational framework for integrating time-series transcriptomics with GRN inference, which can then be used to study the long-term behaviour of the CLL cells through dynamical modelling.

Temporal lobe epilepsy (TLE) is the most prevalent type of focal epilepsy. Recent developments in sequencing, proteomics and network analysis tools provide new avenues for investigating potential molecular therapeutic targets. Both the TGF-β/SMAD signaling pathways and subsets of microRNAs (including miR-21a-5p, miR-142a-5p, and miR-10a-5p) have been shown to be altered in several preclinical models of epilepsy and were mathematically modeled in this study. Using prior systems-based findings, a changeover between 'seizure' and 'anti-seizure' cellular states has been identified upon inhibition of microRNA activity achieved by the injection of antagomirs. Methods for seizure suppression were explored under various antagomir dosages as well as the regulatory effect of each microRNA in order to ascertain intracellular responses. Promising antagomir administration strategies were then identified, which may offer new avenues for seizure suppression.

Bio-pacemakers offer a potential alternative to electronic devices, yet their stable implementation at cellular and tissue levels remains unresolved. In this computational study, we aimed to investigate possible effects of the electrotonic interaction between cardiac cells and the spatial distribution of the bio-pacemaker on the initiation and conduction of cardiac pacemaking action potentials to surrounding quiescent cardiac tissues. Simulation results demonstrated that (i) a combination of weak gap junctional electrical coupling among PMs; and (ii) rectified coupling arising from heterotypic gap junction expressions between the PM and ventricle yielded the best stable and robust bio-pacemaker for pacing and driving surrounding ventricular tissue. Furthermore, Isolated or septal placement improved ventricular pacing efficacy. This study adopts a digital health approach, providing an important theoretical foundation for the simulation of new clinical therapies.

Metabolomics data are often generated through different platforms and quantification methods which makes their synthesis and large-scale replication challenging. This study developed an ensemble of importance-weighted autoencoders to perform cross-platform metabolomics imputation between two metabolomics platforms, Metabolon and National Phenome Centre (NPC) at Imperial College, using 979 samples from the Airwave Health Monitoring Study. The generated samples were highly correlated with real values across all metabolites (µ_ρ = 0.61 (0.55-0.67)). The well-imputed subset contained 199 metabolites (22%), capturing ≥ 55% variance (R² ≥ 0.55) with minimal uncertainty (R² variance ≤ 0.025), including 43 metabolites unique to Metabolon. The concordance of associations in 2,971 validation samples between real and imputed metabolites with two clinical outcomes, body mass index (BMI) and C-reactive protein (CRP), were highly correlated (ρ_BMI = 0.93; ρ_CRP = 0.89) with minimal mean difference (BMI µ_Δ = 0.005 (0.04); CRP µ_Δ = 0.005 (0.04)). Similar concordance occurred with equivalent UK Biobank (BMI µ_Δ = -0.007 (0.05); CRP µ_Δ = 0.01 (0.04)) and NPC (BMI µ_Δ = -0.013 (0.04); CRP µ_Δ = -0.019 (0.04)) metabolites. This methodological innovation offers a scalable and accurate method for cross-platform imputation, enabling the aggregation of metabolomics data from different epidemiological studies for replication and meta-analyses.

Chemical reaction networks (CRNs) are broadly used to study biological systems via simulations. Gillespie's Stochastic Simulation Algorithm (SSA) is commonly used to perform stochastic simulations with CRNs. Comparing two CRNs in such a setting relies on ad hoc signals obtained from the time series, which the simulations output by discarding causal patterns. To this end, we introduce a general method and its implementation for quantitatively comparing CRNs' dynamic behaviour based on causal dependencies in stochastic simulations. Our method detects causal patterns, as in Petri nets, as resource dependencies between reactions during simulation. We present our method within a conservative extension of SSA that tracks and logs these dependencies in simulations as weighted directed graphs. These graphs provide discrete structures that quantify the CRNs' stochastic dynamic behaviour, complementing the simulations' time series output. We use these graphs to compare the behaviour of any two CRNs for the resource dependencies of their components for any time interval. We measure the similarity of the two CRNs via a distance metric. We illustrate different use cases of our method on models of various molecular mechanisms, including gene regulation and drug metabolism.

Cancer therapies often fail when intolerable toxicity or drug-resistant cancer cells undermine otherwise effective treatment strategies. Over the past decade, adaptive therapy has emerged as a promising approach to postpone emergence of resistance by altering dose timing based on tumor burden thresholds. Despite encouraging results, these protocols often overlook the crucial role of toxicity-induced treatment breaks, which may permit tumor regrowth. Herein, we explore the following question: would incorporating toxicity feedback improve or hinder the efficacy of adaptive therapy? To address this question, we propose a mathematical framework for incorporating toxic feedback into treatment design. We find that the degree of competition between sensitive and resistant populations, along with the growth rate of resistant cells, critically modulates the impact of toxicity feedback on time to progression. Further, our conceptual model identifies circumstances where strategic treatment breaks, which may be based on either tumor size or toxicity, can mitigate overtreatment and extend time to progression, both at the baseline parameterization and across a heterogeneous virtual population. Taken together, these findings highlight the importance of integrating toxicity considerations into the design of adaptive therapy.

Chemical space exploration is an important part of chemistry and biology, enabling the discovery and optimization of metabolic pathways, advancing synthetic metabolic functions, and understanding biochemical network evolution. We use a graph-based computational approach implemented in the cheminformatics software MØD, integrated with Integer Linear Programming (ILP) optimization, to systematically search chemical spaces. This approach allows for flexible and targeted queries, including identification of autocatalytic cycles, thermodynamic considerations, and discovery of novel enzymatic cascades. Specifically, we explore the chemical space of natural and artificial carbon fixation pathways defined from relevant enzyme reactions. By applying different optimization criteria, we identify new varieties and recombinations of natural autocatalytic pathways, and compare the properties of the pathways. This work highlights the versatility of graph-based cheminformatics for the purpose of chemical space exploration and artificial pathway design. Potential applications of this framework extend to carbon capture technologies, improved agricultural yields, and value-added chemical production, advancing efforts to address global sustainability challenges.