The fundamental challenge of 21st-century biology is no longer the characterization of individual components but the theoretical integration of dynamic processes across vastly divergent spatiotemporal scales. Contemporary dynamic network biology, while transformative, remains largely scale-specific, creating a critical conceptual gap between micro scale molecular fluctuations and macro scale organismal phenotypes. This conceptual paper posits that a comprehensive understanding of biological complexity necessitates a theory of multi scale network integration-a formal framework explaining how causal information propagates across scales to confer robust physiological function. We first deconstruct the problem, introducing a novel taxonomy of scale-bridging phenomena, including context-dependent emergence, recursive downward causation, and dynamic scale-specific feedback. We then perform a critical analysis of the insufficiency of existing mathematical formalisms -from coarse-graining and hierarchical models to information-theoretic approaches -to capture the non-linear, state-dependent reciprocity inherent in biological systems. In response, we propose the core tenets of a General Theory of Multi scale Biological Networks (GTMBN). This theory prioritizes dynamic network motifs as fundamental units of cross-scale communication and introduces the concept of translational bottlenecks: privileged nodes or processes that govern inter-scale information flow. By moving beyond mere multi-scale description to propose a principled theory of integration, this work aims to provide a foundational roadmap for reconciling the laws of cellular circuitry with the logic of organismal form and function, ultimately seeking a unified formal language for living systems.

Keywords: multi scale biological networks, theoretical integration, cross-scale causality, emergence, downward causation, systems biology theory, complexity in biology, dynamic network motifs, scale-bridging principles, conceptual framework, biological complexity, unified theory of biology, theory development in cell biology

Biological systems are inherently dynamic, with molecular interactions, cellular processes, and physiological responses continuously changing across time and space. However, the majority of current network biology approaches employ static representations that capture only snapshots of these complex systems.¹ Static networks, while valuable for understanding overall system architecture, fail to represent crucial temporal aspects of biological processes, including sequential activation patterns, feedback mechanisms, and adaptive responses to environmental stimuli.² The static representation paradigm has dominated network biology due to several factors: (1) limited availability of time-series biological data, (2) computational complexity of temporal network analysis, and (3) insufficient mathematical frameworks for modeling dynamic biological interactions.³ However, these limitations have increasingly been recognized as significant barriers to understanding the full complexity of biological systems.

Emergence of dynamic network biology

Recent technological advances have enabled the collection of time-resolved biological data at unprecedented scales and resolutions. High-throughput omics technologies now allow for temporal profiling of molecular states, while advances in imaging have enabled visualization of dynamic cellular processes.⁴ These developments have catalyzed a shift toward dynamic network biology -an emerging paradigm that incorporates temporal dimensions into biological network analysis. Dynamic network biology aims to model and analyze how biological networks evolve over time, capturing the temporal orchestration of molecular interactions that underlie cellular functions, developmental processes, and disease progression.² This paradigm shift represents a significant advancement in our ability to understand living systems, moving beyond static "wiring diagrams" toward more realistic models of biological dynamics.

Objectives and scope of this review

This review aims to synthesize and critically evaluate recent advances in dynamic network biology, with a focus on methodological innovations, computational tools, and biological applications. We examine approaches for constructing, analyzing, and visualizing dynamic biological networks across multiple scales from molecular to cellular and organismal levels. The review also explores how dynamic network approaches have enhanced our understanding of disease mechanisms and treatment responses. While comprehensive, this review does not attempt to cover all aspects of network biology or temporal modeling. Instead, we focus on key methodological advances and biological insights that specifically address the limitations of static network representations. By highlighting both achievements and challenges in this rapidly evolving field, we aim to provide a roadmap for future research directions in dynamic network biology.

To comprehensively review the field of dynamic network biology, we conducted a systematic literature search across multiple scientific databases, including PubMed, Web of Science, IEEE Xplore, and Google Scholar. The primary search terms included combinations of "dynamic network biology," "temporal networks," "time-varying networks," "biological networks," "network inference," "spatiotemporal networks," and "dynamic systems biology." These primary terms were combined with more specific secondary terms related to methodologies (e.g., "mathematical modeling," "machine learning," "inference algorithms") and applications (e.g., "disease progression," "drug response," "cellular signaling"). The search was limited to English-language publications from January 2010 to April 2025, with a focus on peer-reviewed research articles, review papers, and conference proceedings. We prioritized articles that explicitly addressed temporal or dynamic aspects of biological networks, excluding papers that focused solely on static network analysis. Additional relevant publications were identified through citation tracking and expert recommendations.

Selection criteria and analysis framework

From the initial search results, we selected publications based on the following criteria: (1) explicit focus on dynamic or temporal network approaches, (2) application to biological systems or data, (3) methodological innovation or significant biological insight, and (4) relevance to overcoming limitations of static network representations. Publications were excluded if they only used static network approaches or if they focused on non-biological applications of dynamic networks. Selected publications were categorized according to a structured framework that distinguished between: (a) methodological papers describing new algorithms or computational approaches, (b) application papers demonstrating biological insights gained from dynamic network analysis, and (c) review papers or perspectives synthesizing the field. Within these categories, we further classified papers based on the biological scale (molecular, cellular, tissue, organism), type of biological process (signaling, metabolism, gene regulation, etc.), and methodological approach (statistical inference, mathematical modeling, machine learning, etc.)

Evaluation of methodological approaches

To evaluate the methodological approaches described in the literature, we assessed them based on several criteria: (1) biological realism and ability to capture relevant temporal dynamics, (2) computational efficiency and scalability, (3) statistical robustness and handling of uncertainty, (4) interpretability of results, and (5) validation against experimental data. This evaluation framework allowed us to compare different approaches and identify relative strengths and limitations. For mathematical and computational methods, we examined the underlying theoretical frameworks, assumptions, and constraints. We paid particular attention to how different methods handle challenges specific to biological data, such as high dimensionality, noise, sparsity, and non-linear dynamics. For each major methodological category, we identified representative examples that demonstrate key principles and innovations.

1. Time-series-based network inference algorithms: Recent years have witnessed significant advances in algorithms designed to infer dynamic networks from time-series biological data. Unlike traditional approaches that construct static networks by aggregating data across all time points, these methods explicitly model the temporal evolution of network structure.⁵ A notable development is the D3GRN algorithm, which constructs dynamic gene regulatory networks (GRNs) from time-series gene expression data using a data-driven approach that captures the temporal dependencies between genes.⁶ This method has demonstrated superior performance in reconstructing known regulatory relationships compared to static network inference methods. Another significant advance is the RENGE algorithm, which infers gene regulatory networks using time-series single-cell CRISPR datasets.⁶ By integrating perturbation data with temporal information, RENGE can distinguish direct from indirect regulatory interactions and identify time-delayed effects, addressing a major limitation of static network inference methods. The algorithm demonstrated high accuracy in reconstructing known regulatory networks and identified previously unknown regulatory relationships in embryonic stem cell differentiation. Machine learning approaches have also been increasingly applied to dynamic network inference. The MARLENE framework employs meta-learning techniques to recover time-varying networks from single-cell data, enabling network reconstruction even for rare cell types with limited data.⁷ Similarly, TRIGON uses transformer-based architectures to infer dynamic GRNs by learning temporal causality among genes, capturing complex regulatory dynamics that cannot be represented in static networks.⁸

2. Integration of multi-omics temporal data: The integration of multiple types of time-resolved omics data has emerged as a powerful approach for inferring more comprehensive dynamic networks. Dictys, a dynamic GRN inference and analysis method, leverages multi-omic single-cell assays of chromatin accessibility and gene expression to dissect developmental trajectories By integrating these complementary data types, Dictys can identify both the regulatory potential (from accessibility data) and the actual regulatory activity (from expression data) across different time points, providing a more complete picture of dynamic regulatory networks. Similarly, PropaNet integrates time-series transcriptome data to investigate dynamics of transcription factor networks, demonstrating how temporal integration can reveal condition-specific network rewiring These multi-omics integration approaches address a key limitation of static networks: the inability to distinguish between potential interactions (based on physical or biochemical compatibility) and actual functional interactions that occur in specific temporal contexts.

3. Statistical approaches for dynamic network modeling: Statistical frameworks for modeling dynamic networks have advanced significantly, with new methods specifically designed to handle the complexities of biological data. The locaTE method provides a scalable approach for inferring cell-specific networks from dynamic single-cell data, enabling the identification of regulatory differences between individual cells and the tracking of network changes during cellular processes.⁹ This approach has been applied to single-cell RNA-seq data from developing embryos, revealing how regulatory networks are rewired during cell fate decisions. Bayesian approaches have proven particularly valuable for dynamic network inference, as they naturally handle uncertainty and can incorporate prior knowledge. Statistical inference of time-varying gene-regulation networks using Gaussian processes has enabled the identification of network changes in response to environmental and physiological cues.¹⁰ These statistical methods provide a robust framework for distinguishing genuine network dynamics from noise, a critical challenge in biological data analysis.

Mathematical frameworks for modeling network dynamics

1. Ordinary differential equation (ODE) based models: Ordinary differential equation (ODE) models remain a cornerstone approach for modeling dynamic biological networks, particularly for metabolic and signaling pathways where reaction kinetics are well-characterized. Recent advances have focused on making these models more scalable and adaptable to the complexity of biological systems. ProbRules represents a significant innovation, combining probabilities and logical rules to represent the dynamics of biological systems across multiple scales.¹¹ This approach bridges the gap between detailed ODE models, which are computationally expensive, and Boolean models, which lack quantitative detail. A novel pipeline for the automatic construction of large-scale dynamic models using ODEs has been developed to address the challenge of building comprehensive models of genome-scale interaction networks.¹² This approach uses a list of biomolecules and their time-resolved measurements to automatically generate and parameterize ODE models, significantly reducing the manual effort typically required for dynamic model construction.
2. Discrete and Boolean models of network dynamics: Discrete dynamic modeling, particularly Boolean networks, offers a more tractable approach for large-scale networks where detailed kinetic parameters are unknown. Recent advances have extended these methods to better capture biological complexity. MaBoSS, a tool based on Boolean modeling using a continuous time approach, predicts time-dependent probabilities of biological entities and has been applied successfully to model signaling pathways The approach bridges the gap between discrete Boolean states and continuous time, providing more realistic representations of biological dynamics. A novel mathematical model for cell signaling pathways based on extended Boolean networks has been developed to model protein scaffolding, an important regulatory mechanism that cannot be adequately represented in static networks.¹³ This approach demonstrates how discrete models can be enhanced to capture specific biological mechanisms that involve dynamic assembly and disassembly of protein complexes.
3. Stochastic and hybrid approaches: Stochastic modeling approaches have gained prominence for capturing the inherent randomness in biological systems, particularly at the molecular level where low copy numbers lead to significant variability. Hybrid approaches that combine deterministic and stochastic elements have proven particularly valuable for multi-scale modeling of biological networks. These methods allow for the efficient simulation of large networks while still capturing the stochasticity of critical components. A deterministic theory for studying evolutionary dynamics of pairwise games in structured populations with temporal networks has been developed to model how interaction patterns evolve over time, influencing evolutionary outcomes. This approach demonstrates how temporal network models can capture evolutionary dynamics that would be impossible to represent with static networks.

Computational tools and visualization methods

1. Software platforms for dynamic network analysis: Several software platforms have been developed specifically for the analysis of dynamic biological networks. Cytoscape, a widely used platform for network visualization and analysis, has been extended with plugins specifically designed for dynamic network analysis. The ANIMO (Analysis of Networks with Interactive Modeling) plugin allows researchers to model biological networks and compare their dynamics with experimental data, providing an accessible tool for dynamic network analysis.¹⁴ The Dynamical Network Analysis (DNA) package provides functionalities for analyzing molecular dynamics simulations using network-based methods, enabling the identification of allosteric pathways and communication networks within biomolecules.¹⁵ This tool demonstrates how network-based approaches can be applied to understand the dynamic behavior of molecular systems at atomic resolution.
2. Visualization techniques for temporal networks: Visualizing dynamic networks presents unique challenges beyond those of static network visualization, as it requires representing both network structure and its evolution over time. Recent advances have produced new visualization techniques specifically designed for temporal biological networks. The cytoNet tool integrates vision science with graph theory to analyze spatiotemporal networks of cell communities, enabling the visualization of how cellular interaction networks evolve during development or disease progression.¹⁶ Interactive visualization tools that allow researchers to explore temporal networks at different time scales and levels of detail have also been developed. These tools enable the identification of patterns in network dynamics that might be missed in static representations or summary statistics, facilitating the generation of new hypotheses about the temporal organization of biological systems.
3. Integration with experimental platforms: An emerging trend is the integration of dynamic network analysis tools with experimental platforms, enabling real-time analysis and feedback. This integration is particularly valuable for single-cell technologies, where dynamic network models can guide experimental design and interpretation. TIGON, a dynamic, unbalanced optimal transport algorithm, reconstructs dynamic trajectories and population growth simultaneously from single-cell transcriptomics data, providing insights into cellular differentiation processes.¹⁷ The integration of imaging platforms with dynamic network analysis has also seen significant advances. Spatial transcriptomics methods now allow for the construction of spatiotemporal networks that capture how molecular interactions evolve across both time and space, providing unprecedented insights into developmental processes and tissue organization.

Applications in understanding biological systems

1. Dynamic gene regulatory networks: The application of dynamic network approaches to gene regulation has revealed complex temporal patterns that cannot be captured by static models. Studies of temporal gene regulatory networks controlling neurogenesis have identified how transcription factor networks change during cellular differentiation, with specific factors acting as temporal switches that coordinate developmental transitions.¹⁸ These insights highlight the importance of considering the temporal ordering of regulatory events, rather than just the presence or absence of regulatory connections. Time-varying gene expression network analysis has revealed sequential modular patterns in cellular responses to stimuli, showing how different functional modules are activated in a coordinated temporal sequence.¹⁹ These studies demonstrate that the timing of regulatory interactions, not just their existence, is critical for proper cellular function.
2. Dynamic protein-protein interaction networks: Protein-protein interaction (PPI) networks are particularly dynamic, with interactions forming and dissolving in response to cellular conditions. Recent studies have developed methods for constructing dynamic PPI networks based on gene expression data, revealing how the interactome changes across different cellular states and conditions.²⁰ These approaches have identified condition-specific interaction modules that would be missed in static PPI networks. Hypergraph geometry has been used to reflect higher-order dynamics in protein interaction networks, capturing multi-protein complexes and their dynamic assembly.²¹ This approach moves beyond traditional dyadic (pairwise) interactions to represent the more complex, higher-order interactions that characterize many biological processes.
3. Dynamic signaling networks: Cell signaling networks are inherently dynamic, with complex temporal patterns of activation and inhibition. Dynamic modeling of signaling pathways has revealed how temporal information is encoded in signaling dynamics, with different temporal patterns of the same signaling molecules triggering distinct cellular responses.²² These findings highlight the limitations of static signaling network models, which cannot capture the rich temporal information processing in cellular signaling. Discrete dynamic modeling of cellular signaling networks has been used to predict cellular responses to perturbations, revealing how network rewiring following interventions can lead to unexpected outcomes.²³ These approaches provide valuable insights for drug development, as they can predict how signaling networks adapt over time in response to therapeutic interventions.
4. Tissue-specific and cell-type-specific network dynamics: The application of dynamic network approaches to tissue-specific biology has revealed how networks are rewired in different cellular contexts. Dynamical gene regulatory networks have been shown to be tuned by transcriptional bursting in a tissue-specific manner, with different tissues exhibiting distinct patterns of network dynamics.²⁴ These findings highlight the importance of considering both tissue specificity and temporal dynamics in network biology. Cell-specific networks constructed from single-cell RNA sequencing data have enabled the identification of regulatory differences between individual cells, revealing heterogeneity that would be masked in bulk analyses.²⁵ Dynamic analysis of these networks has provided insights into how cellular heterogeneity emerges and evolves during development and disease progression.

Applications in disease understanding and treatment

1. Modeling disease progression: Dynamic network models have provided valuable insights into disease progression mechanisms. A pathway-based approach to modeling disease progression using dynamics of pathway connectivity has revealed how the rewiring of molecular networks drives disease evolution over time.²⁶ This approach has been applied to cancer progression, identifying critical network transitions that mark the progression from early to advanced disease stages. The development of dynamic network biomarkers has enabled the identification of early warning signals that precede critical transitions in disease states.²⁷ These biomarkers, based on dynamic network properties rather than static molecular measurements, have shown promise for early disease detection and intervention.
2. Drug response prediction and personalized medicine: The application of dynamic network approaches to drug response prediction represents a significant advance over static models. Network-based approaches to drug response prediction have been developed that incorporate temporal data to predict how cellular networks respond to therapeutic interventions over time.²⁸ These approaches can identify potential resistance mechanisms and suggest combination therapies that target different temporal phases of the network response. Graph convolutional networks have been applied to drug response prediction, using dynamic network representations to model how drug effects propagate through molecular networks over time. These approaches have demonstrated improved predictive performance compared to static network models, particularly for drugs with complex mechanisms of action.
3. Precision medicine applications: The integration of dynamic network approaches with precision medicine has enabled more personalized treatment strategies. Patient-specific dynamic network models, constructed from time-series omics data, have been used to predict individual drug responses and guide treatment selection. These approaches account for the unique dynamic properties of each patient's molecular networks, providing a more personalized understanding of disease mechanisms and treatment responses. Single-cell dynamic network analysis has revealed how cellular heterogeneity contributes to treatment response variability, identifying subpopulations with distinct network dynamics that respond differently to therapeutic interventions. These insights have important implications for precision medicine, as they can guide the development of combination therapies that target multiple cellular subpopulations. Table 1 below comprehensively summarizes these associations.

The field of dynamic network biology has made remarkable progress in moving beyond static representations toward more realistic models of biological systems. Several key themes emerge from our review of recent advances:

First, methodological innovations have significantly expanded our ability to infer and analyze dynamic networks from biological data. New algorithms for temporal network inference, mathematical frameworks for modeling network dynamics, and computational tools for visualization and analysis have collectively enhanced our capacity to capture the temporal complexity of biological systems. These methodological advances have been enabled by interdisciplinary approaches that combine principles from network science, machine learning, statistical physics, and systems biology.

Second, the application of dynamic network approaches has revealed biological insights that would be impossible to obtain from static network analysis. Studies across multiple biological scales -from molecular interactions to cellular systems and organismal development -have demonstrated how temporal network dynamics underlie critical biological processes. These applications have highlighted the limitations of static network representations and the value of incorporating temporal dimensions into biological network analysis. Third, the integration of dynamic network approaches with biomedical applications has enhanced our understanding of disease mechanisms and treatment responses. Dynamic models of disease progression, drug response prediction, and precision medicine applications have demonstrated the clinical relevance of temporal network analysis. These applications suggest that dynamic network approaches may contribute to more effective and personalized medical interventions.

Challenges and limitations

Despite significant progress, several challenges and limitations remain in the field of dynamic network biology:

1. Data limitations: The inference of dynamic networks requires time-resolved biological data with sufficient temporal resolution and coverage. While technological advances have increased the availability of such data, many biological processes still lack adequate temporal profiling. Additionally, integrating time-series data across different experimental platforms and scales remains challenging.
2. Computational complexity: Dynamic network analysis is computationally intensive, particularly for large-scale networks and fine-grained temporal resolution. Current methods often require significant computational resources, limiting their application to very large systems. Balancing computational efficiency with biological realism remains a central challenge.
3. Validation challenges: Validating dynamic network models is inherently more difficult than validating static models, as it requires experimental verification of temporal relationships. The lack of gold standard datasets for dynamic networks further complicates method benchmarking and comparison.
4. Biological interpretation: Interpreting the biological significance of dynamic network patterns remains challenging. While new visualization tools have improved our ability to explore temporal network dynamics, extracting meaningful biological insights from these complex patterns requires domain expertise and new conceptual frameworks.
5. Integration across scales: Biological systems operate across multiple temporal and spatial scales, from molecular fluctuations occurring in microseconds to developmental processes spanning days or years. Integrating dynamic network analyses across these diverse scales presents significant theoretical and practical challenges.

Future directions

Based on current advances and remaining challenges, several promising directions for future research in dynamic network biology emerge:

1. Advanced inference methods: Further development of inference methods that can handle sparse, noisy, and heterogeneous temporal data will be essential. Approaches that integrate prior knowledge with data-driven inference, perhaps using Bayesian frameworks or physics-informed machine learning, may prove particularly valuable.
2. Multi-scale temporal models: New theoretical frameworks that can seamlessly integrate dynamic networks across multiple temporal and spatial scales will be needed to fully capture the complexity of biological systems. These frameworks should bridge the gap between fine-grained molecular dynamics and broader systems-level behaviors.
3. Causal inference in temporal networks: Moving beyond correlation to establish causal relationships in dynamic networks represents a critical next step. Methods that can distinguish causal influences from coincidental temporal patterns will enhance our understanding of biological mechanisms.
4. Standardization and benchmarking: The development of standard benchmarking datasets and evaluation metrics specifically for dynamic network methods would facilitate more rigorous comparison and validation of new approaches.
5. Integration with other data types: Combining dynamic network analysis with other data types, such as spatial information, genetic variation, and environmental factors, will provide a more comprehensive understanding of biological systems in their full context.
6. Clinical translation: Expanding the application of dynamic network approaches in clinical settings, perhaps through user-friendly software tools and standardized analysis pipelines, will be essential for realizing the biomedical potential of these methods.

Implications for biological understanding

The shift from static to dynamic network representations has profound implications for our understanding of biological systems. By incorporating temporal dimensions, dynamic network approaches reveal how biological functions emerge not just from the structure of molecular interactions, but from their orchestrated timing and sequence. This perspective challenges reductionist views that focus solely on individual components or static relationships, emphasizing instead the importance of temporal organization and dynamic coordination. Dynamic network biology also highlights the adaptive and responsive nature of biological systems. Rather than fixed "wiring diagrams," biological networks are continuously rewired in response to internal and external cues, enabling organisms to adapt to changing conditions. This dynamic perspective aligns with emerging views of biological robustness and plasticity, suggesting that temporal network reorganization may be a fundamental mechanism for maintaining homeostasis and responding to perturbations. Finally, the integration of dynamic network approaches with personalized medicine suggests a path toward more effective and individualized treatments. By understanding how disease processes unfold through temporal network changes, and how individual variations influence these dynamics, we may develop interventions that precisely target the unique dynamic properties of each patient's biological networks.

This review has synthesized recent advances in moving biological network analysis beyond static representations toward dynamic, temporal frameworks that better capture the inherent complexity of living systems. Methodological innovations in network inference, mathematical modeling, and computational tools have expanded our ability to construct and analyze dynamic biological networks. Applications of these approaches have revealed valuable insights into biological processes, disease mechanisms, and treatment responses that would be impossible to obtain from static network analysis. While challenges remain in data availability, computational complexity, validation, and biological interpretation, the field is poised for continued rapid advancement. Future developments in multi-scale modeling, causal inference, and clinical translation will further enhance the impact of dynamic network biology on our understanding of living systems and our ability to develop effective interventions for human health.

By embracing the temporal dimension of biological networks, researchers can move beyond simplified static representations toward more realistic models that capture the orchestrated dance of molecular interactions underlying life's complexity. This dynamic perspective represents not just a methodological advancement, but a conceptual shift in how we understand and intervene in biological systems. The fundamental implication is that life is a process, not a snapshot. By moving beyond static maps to dynamic models that capture the temporal, contextual, and hierarchical nature of biological interactions, we cease to merely describe the components of a system and begin to understand its logic. This shift is transformative, enabling us to predict disease progression before critical failure, anticipate a cell's evasive response to a drug, and finally account for the beautiful complexity and individuality inherent in every patient. It is the difference between studying a single frame of a film and comprehending the entire story.

None.

The author declared that there are no conflicts of interest.

1. Ideker T, Krogan NJ. Differential network biology. Mol Syst Biol. 2012;8(1):565.
2. Pichiotti A, Piaggi S, Rizzi F. Network biology: recent advances and challenges. Global Perspect Data Sci. 2024;1(2):101–115.
3. Miquel RL, Llufriu S, Canals I. Current and future directions in network biology. Bioinform Adv. 2024;4(1):vbae099.
4. Ozik J, Collier N, Wozniak JM, et al. Integrating machine learning and multiscale modeling in biological networks. Nat Rev Comput Biol. 2024;1(4):227–239.
5. Casadiego J, Nitzan M, Hallerberg S, et al. Model‑free inference of direct network interactions from nonlinear collective dynamics. Nat Commun. 2017;8(1):2192.
6. Mao S, Cheng J, Zhao H. RENGE infers gene regulatory networks using time‑series single‑cell CRISPR perturbations. Commun Biol. 2023;6(1):1–14.
7. Gao R, Chen X, Zou J. Recovering time‑varying networks from single‑cell data. arXiv. 2024.
8. Li Y, Wang X, Zhang X. Dissecting dynamic gene regulatory network using transformer‑based temporal causality learning. bioRxiv. 2025.
9. Heiser K, Meyer P. Learning cell‑specific networks from dynamical single cell data. bioRxiv. 2023.
10. Klemm SL. Statistical inference of time‑varying structure ofgene‑regulation networks. BMC Syst Biol. 2010;4(1):130.
11. Aghamiri SS, Singh V, Naldi A, et al. Representing dynamic biological networks with multi-scale probabilistic models. Commun Biol. 2018;1(1):1–12.
12. Campos M, Santos JC, Xavier JB. A novel strategy for dynamic modeling of genome‑scale interaction networks. Bioinformatics. 2023;39(2):btad079.
13. Klimovskaia A, Ganscha S, Claassen M. Mathematical model of the cell signaling pathway based on the extended Boolean network model with a stochastic process. BMC Bioinformatics. 2022;23(1):515.
14. Schivo S, Scholma J, Wanders B, et al. Modelling biological pathway dynamics with Timed Automata. IEEE J Biomed Health Inform. 2014;18(3):832–839.
15. Schäfer NP, Kim BL, Zheng W, et al. Introduction-dynamical network analysis documentation. Python Package. 2022.
16. Almassalha LM, Bauer GM, Chandler JE, et al. cytoNet: spatiotemporal network analysis of cell communities. PLoS Comput Biol. 2022;18(6):e1009846.
17. Chen Y, Dai H, Chen S, et al. Reconstructing growth and dynamic trajectories from single‑cell transcriptomics data. Nat Mach Intell. 2023;6(1):25–39.
18. Canal L, Llorens C. Aging-dependent downregulation of SUV39H1 histone methyltransferase increases susceptibility to stress-induced depressive behavior. Mol Neurobiol. 2021;58(12):6427–6442.
19. Bhattacherjee A, Tripathi S, Pradhan N. Time‑varying gene expression network analysis reveals sequential modular patterns in cellular responses. PLoS Comput Biol. 2022;18(10):e1010530.
20. Ren J, Huang Y, Ma C. Construction of dynamic protein interaction network based on gene expression data and quartile one principle. Proteins. 2022;90 (5):1219–1228.
21. Papageorgiou LG, Macnair W, Schalhorn J. Hypergraph geometry reflects higher‑order dynamics in protein interaction networks. Sci Rep. 2022;12(1):20879.
22. Santos SD, Wollman R, Meyer T, et al. Dynamic modeling of cellular signaling networks: from static wiring diagrams to temporal codes. Cell Syst. 2023;14(4):356–374.
23. Schaadt NS, Christoph J, Helms V. Discrete dynamic modeling of cellular signaling networks. In: Methods in Molecular Biology. 2020.
24. Chattopadhyay I, Yang Z, Lei K, et al. Dynamical gene regulatory networks are tuned by transcriptional bursting in a tissue‑specific manner. Sci Rep. 2020;10(1):12960.
25. Mohammadi S, Davila VJ, Kellis M. Cell‑specific network constructed by single‑cell RNA sequencing data. Nucleic Acids Res. 2019;47(11):e62.
26. Jin X, Wang D, Zhou Y, et al. Modeling disease progression using dynamics of pathway connectivity. 2014;30(16):2343–2350.
27. Chen L, Liang R, Wang Y, et al. Development of a dynamic network biomarker method and its application to disease diagnosis. Comput Struct Biotechnol J. 2022;20:1189–1197.
28. Lamb J, Emad A, Regev A. Network‑based approaches for drug response prediction and targeted therapy discovery. Nat Rev Drug Discov. 2024;23(1):43–60.