Metabolic network topological analysis is a research field at the intersection of systems biology, complex network theory, and statistical modeling. It represents metabolic systems as networks composed of nodes and edges, and uses network structure to describe the relationships among metabolites, metabolic reactions, pathways, organs, or disease phenotypes. Nodes may represent metabolites, reactions, enzymes, pathways, or organs, while edges may be derived from known biochemical reactions or from correlations, partial correlations, differential correlations, coordinated changes, or functional associations inferred from high-throughput data. Unlike analyses that focus only on changes in individual metabolite abundance, metabolic network topological analysis asks who is connected to whom, which nodes occupy key positions, which modules are reorganized, and how disease, drug treatment, or environmental perturbation reshapes the overall network structure.

Historically, metabolic network topological analysis originated from the application of complex network science and systems biology to the study of cellular functional organization. In 2000, Jeong and colleagues systematically compared the metabolic networks of 43 organisms and found that metabolic networks display clear non-random organizational properties, indicating that metabolic systems are not simple collections of isolated reactions but complex networks with intrinsic structural principles [1]. Ravasz and colleagues subsequently revealed the hierarchical modular organization of metabolic networks, showing that local functional modules can be progressively assembled into larger system-level structures [2]. Barabási and Oltvai later provided a systematic framework for network biology, helping establish complex network analysis as an important tool for understanding cellular function, disease mechanisms, and system homeostasis [3]. Since then, metabolic network research has gradually moved from static reaction maps toward omics-driven state-specific networks, differential networks, multiscale networks, and individualized network analysis.

The basic metrics of metabolic network topological analysis include degree, clustering coefficient, betweenness centrality, closeness centrality, modularity, shortest path length, network density, and network connectivity. Degree reflects the number of connections associated with a node. Betweenness centrality reflects the role of a node in linking different regions of the network. The clustering coefficient describes the local connectedness among neighboring nodes. Modularity indicates whether the network contains relatively independent functional units. In metabolic systems, these metrics are not merely abstract mathematical quantities. They help researchers determine which metabolites may act as hubs, which pathways may be jointly perturbed, and which local structures are rewired under disease conditions.

A key feature of this field is its relationship-oriented perspective. Traditional differential analysis usually answers the question of which metabolites have changed, whereas metabolic network topological analysis further asks how these changes are related to one another. Some metabolites may show no significant change in abundance, yet their connections with other metabolites may change markedly. Such nodes may participate in regulatory rewiring under disease states and are often difficult to identify through abundance-based screening alone. Network topological analysis can capture this type of system-level perturbation reflected by changes in connection patterns.

Another important feature of metabolic network topological analysis is its emphasis on dynamic rewiring. Metabolic associations that remain stable under healthy conditions may be weakened, strengthened, or replaced during disease progression, drug intervention, environmental exposure, or aging. Hub nodes may shift, original modules may split, and new connections may emerge between modules. This type of network rewiring is often closer to system-level pathophysiology than changes in individual metabolites. For complex diseases, the central question is not only whether a molecule increases or decreases, but whether the connection pattern of the metabolic system has been reorganized.

Metabolic network topological analysis also has a multiscale nature. Metabolic networks can be constructed at the metabolite level and can also be extended to the levels of pathways, tissues, organs, and diseases. Metabolite-level networks are suitable for analyzing local metabolic coordination. Pathway-level networks help explain relationships among functional modules. Organ-level networks can describe systemic metabolic responses. Disease-level networks can be used to study comorbidity relationships. The value of multiscale analysis lies in its ability to link local metabolic perturbations with system-level functional changes, allowing researchers to move from single-level observation toward cross-level interpretation.

The significance of metabolic network topological analysis lies in its ability to provide a systems-level explanatory framework for complex diseases. In tumors, neurodegenerative diseases, diabetic nephropathy, liver diseases, and inflammation-related disorders, metabolic abnormalities are often not caused by changes in a single molecule, but involve multiple pathways, tissues, and regulatory processes. Network topological analysis can help identify key nodes, key edges, perturbed modules, and potential propagation routes, providing new clues for disease subtyping, mechanistic studies, biomarker panel selection, and intervention target discovery. The work of Lee and colleagues on human metabolic network topology and disease comorbidity also suggests that network structure can provide important explanations for relationships among diseases [6].

This field also faces several methodological challenges. First, the way a network is constructed directly affects the interpretation of the results. Networks based on biochemical reactions, correlations, partial correlations, and differential associations do not carry the same meaning. Second, the reliability of edges must be carefully controlled. Limited sample size, noise, batch effects, and threshold selection may all introduce spurious connections. Third, topological metrics have mathematical meaning, but they do not necessarily imply biological causality. A node with high centrality can be considered structurally important in the network, but its biological role still requires validation using experiments, independent cohorts, or external knowledge. Therefore, metabolic network topological analysis must take statistical robustness, biological interpretability, and result validation into account at the same time.

Our team's research interests in metabolic network topological analysis mainly focus on metabolic network construction, topological perturbation quantification, multiscale synergy networks, differential correlation analysis, and individualized network rewiring. The team has long worked on molecular network modeling, machine learning, spatial multi-omics, and health big data analysis. We aim to explain systemic metabolic abnormalities in complex diseases by examining coordinated relationships among metabolites, functional associations among pathways, and network rewiring under disease states. Compared with conventional metabolite screening, our work places greater emphasis on the connection structure of metabolic systems and how these connections change during disease onset and progression.

In information entropy and topological effect quantification, our team focuses on how to measure local network structural changes under disease conditions. Traditional network metrics can describe the importance of nodes, but they often cannot directly quantify the contribution of a metabolite, a metabolite pair, or a subnet to overall topological change. To address this issue, our team has been exploring concepts such as TES, or Topology Effect Size, with the aim of integrating information entropy, network perturbation, and statistical testing to establish a metric system capable of characterizing the strength of local topological changes. The goal of these methods is not merely to determine whether an edge exists, but to further quantify how local structural changes affect the global network state.

In metabolite set and pathway network analysis, our team focuses on problems such as pathway overlap, incomplete metabolite coverage, and complex correlation structures. Traditional pathway enrichment analysis often treats pathways as relatively independent sets. In real metabolic systems, however, different pathways share many metabolites and are strongly interconnected. The overlapping group PLS method proposed by our team incorporates overlapping metabolite sets into the modeling process and improves the ability of metabolite set analysis to represent pathway intersection structures [8]. The subsequently developed iMSEA method further incorporates metabolite association networks and pathway structures to interpret the metabolic mechanisms underlying drug interactions from a network perspective [10]. These studies reflect our team’s transition from metabolite lists toward pathway relationship modeling.

In differential correlation-driven analysis of metabolic heterogeneity, our team focuses on whether the coordination relationships among metabolites are altered under disease states. The dci-MSEA method introduces differential correlation information into metabolite set enrichment analysis to identify disease phenotype-associated metabolic network heterogeneity [11]. The central idea of this type of method is that disease may alter not only metabolite concentrations, but also the relationships among metabolites. By comparing connection patterns across different states, it becomes possible to identify perturbed pathways and key subnets that are difficult to detect using conventional abundance-based analysis.

In multiscale and cross-organ metabolic synergy networks, our team has proposed methods such as iMS2Net and SynNet to analyze coordinated relationships in metabolic systems across different levels. iMS2Net is designed for the construction and interpretation of multiscale metabolic synergy networks and can be used to analyze systemic metabolic responses to external stimuli [9]. SynNet further constructs multiscale synergy networks associated with specific metabolic phenotypes and is used to investigate disease and comorbidity mechanisms [12]. These methods emphasize that important information in metabolic systems is not contained only in individual metabolites or single pathways, but also in the connection patterns among metabolites, pathways, tissues, and disease phenotypes.

In individualized network topological analysis, our team focuses on metabolic network rewiring at the single-sample level. Complex diseases show substantial individual variation, and population-level average networks may obscure patient-specific metabolic connection patterns. ssNetShift uses single-sample network estimation and topological rewiring measures to identify individualized topological driver metabolites, revealing disease risk information and prognosis-related subtypes that are difficult to detect using conventional abundance analysis [13]. This direction moves metabolic network topological analysis from group-level comparison toward individualized mechanistic interpretation and provides a new methodological basis for metabolic risk assessment in precision medicine.

In tool development and translational application, our team emphasizes converting network topological analysis methods into reusable software and analytical workflows. Complex disease research requires not only a single topological metric, but also an integrated toolchain covering data preprocessing, network construction, perturbed subnet identification, topological significance testing, and visual interpretation. The team has continued to develop algorithms, software, and patentable methods for perturbed metabolic subnet analysis, metabolite set network modeling, multiscale synergy networks, and individualized network rewiring. These tools support the application of metabolic network topological analysis in cancer, neurodegenerative diseases, metabolic disorders, and studies of drug mechanisms of action.

Overall, our team's goal in metabolic network topological analysis is to develop systems-level metabolic network analysis methods for complex diseases. On the one hand, we focus on constructing stable, reliable, and interpretable metabolic networks while reducing spurious connections caused by noise, threshold selection, and limited sample size. On the other hand, we aim to identify key nodes, key edges, perturbed modules, topological drivers, and disease-associated rewiring patterns from network structures. Through these studies, metabolic network topological analysis can move beyond the interpretation of individual differential metabolites and become a systems-level tool for metabolic mechanism analysis, providing methodological support for complex disease mechanism research, molecular subtyping, comorbidity analysis, and individualized risk assessment.

References

[1] Jeong, H.; Tombor, B.; Albert, R.; Oltvai, Z. N.; Barabási, A.-L. The Large-Scale Organization of Metabolic Networks. Nature, 2000, 407(6804): 651–654.

[2] Ravasz, E.; Somera, A. L.; Mongru, D. A.; Oltvai, Z. N.; Barabási, A.-L. Hierarchical Organization of Modularity in Metabolic Networks. Science, 2002, 297(5586): 1551–1555.

[3] Barabási, A.-L.; Oltvai, Z. N. Network Biology: Understanding the Cell’s Functional Organization. Nature Reviews Genetics, 2004, 5(2): 101–113.

[4] Guimerà, R.; Amaral, L. A. N. Functional Cartography of Complex Metabolic Networks. Nature, 2005, 433(7028): 895–900.

[5] Duarte, N. C.; Becker, S. A.; Jamshidi, N.; Thiele, I.; Mo, M. L.; Vo, T. D.; Srivas, R.; Palsson, B. Ø. Global Reconstruction of the Human Metabolic Network Based on Genomic and Bibliomic Data. Proceedings of the National Academy of Sciences of the United States of America, 2007, 104(6): 1777–1782.

[6] Lee, D.-S.; Park, J.; Kay, K. A.; Christakis, N. A.; Oltvai, Z. N.; Barabási, A.-L. The Implications of Human Metabolic Network Topology for Disease Comorbidity. Proceedings of the National Academy of Sciences of the United States of America, 2008, 105(29): 9880–9885.

[7] Zhao, C.; Dong, J.; Deng, L.; Tan, Y.; Jiang, W.; Cai, Z. Molecular Network Strategy in Multi-Omics and Mass Spectrometry Imaging. Current Opinion in Chemical Biology, 2022, 70: 102199.

[8] Deng, L.; Ma, L.; Cheng, K. K.; Xu, X.; Raftery, D.; Dong, J. A Sparse PLS-Based Method for Overlapping Metabolite Set Enrichment Analysis. Journal of Proteome Research, 2021, 20(6): 3204–3213.

[9] Dong, J.; Peng, Q.; Deng, L.; Liu, J.; Huang, W.; Zhou, X.; Zhao, C.; Cai, Z. iMS2Net: A Multiscale Networking Methodology to Decipher Metabolic Synergy of Organism. iScience, 2022, 25(9): 104896.

[10] Wang, Y.; Liu, X.; Dong, L.; Cheng, K. K.; Lin, C.; Wang, X.; Dong, J.; Deng, L.; Raftery, D. iMSEA: A Novel Metabolite Set Enrichment Analysis Strategy to Decipher Drug Interactions. Analytical Chemistry, 2023, 95(15): 6203–6211.

[11] Lin, G.; Dong, L.; Cheng, K. K.; Xu, X.; Wang, Y.; Deng, L.; Raftery, D.; Dong, J. Differential Correlations Informed Metabolite Set Enrichment Analysis to Decipher Metabolic Heterogeneity of Disease. Analytical Chemistry, 2023, 95(33): 12505–12513.

[12] Wang, Y.; Zhu, Z.; Deng, L.; Cheng, K. K.; Guo, F.; Lin, G.; Raftery, D.; Dong, J. Multiscale Synergy Networks Offer Insights into Disease and Comorbidity Mechanisms. Analytical Chemistry, 2025, 97(6): 3633–3642.

[13] Lin, G.; Li, S.; Cheng, K. K.; Xu, X.; Deng, L.; Dong, J.; Raftery, D. ssNetShift: Single-Sample Metabolic Network Rewiring Reveals Hidden Prognostic Subtypes beyond Clinical Staging in Gastric Cancer. Briefings in Bioinformatics, 2026, 27(3): bbag264.