Register or Submit a manuscript.
Quick Link
OBM Genetics

(ISSN 2577-5790)

OBM Genetics is an international Open Access journal published quarterly online by LIDSEN Publishing Inc. It accepts papers addressing basic and medical aspects of genetics and epigenetics and also ethical, legal and social issues. Coverage includes clinical, developmental, diagnostic, evolutionary, genomic, mitochondrial, molecular, oncological, population and reproductive aspects. It publishes a variety of article types (Original Research, Review, Communication, Opinion, Comment, Conference Report, Technical Note, Book Review, etc.). There is no restriction on the length of the papers and we encourage scientists to publish their results in as much detail as possible.

Publication Speed (median values for papers published in 2024): Submission to First Decision: 6.4 weeks; Submission to Acceptance: 12.2 weeks; Acceptance to Publication: 7 days (1-2 days of FREE language polishing included)

Current Issue: 2025  Archive: 2024 2023 2022 2021 2020 2019 2018 2017
Open Access Original Research

Network Topology Similarities Across Cancer Types: Identifying Central Protein Hubs for Drug Discovery

Emad Fadhal 1,2,*

  1. Department of Mathematics & Statistics, College of Science, King Faisal University, P. O. Box 400, Al-Ahsa 31982, Saudi Arabia

  2. Department of Mathematics and Applied Mathematics, University of the Western Cape, Bellville, South Africa

* Correspondence: Emad Fadhal

Academic Editor: Lunawati L Bennett

Received: June 18, 2025 | Accepted: October 07, 2025 | Published: October 13, 2025

OBM Genetics 2025, Volume 9, Issue 4, doi:10.21926/obm.genet.2504312

Recommended citation: Fadhal E. Network Topology Similarities Across Cancer Types: Identifying Central Protein Hubs for Drug Discovery. OBM Genetics 2025; 9(4): 312; doi:10.21926/obm.genet.2504312.

© 2025 by the authors. This is an open access article distributed under the conditions of the Creative Commons by Attribution License, which permits unrestricted use, distribution, and reproduction in any medium or format, provided the original work is correctly cited.

Abstract

A molecular-level understanding of cancer is essential for the development of effective therapies. Constructing protein-protein interaction (PPI) networks offers a valuable approach to identifying dysregulated driver genes and potential therapeutic targets. In this study, we modeled cancer PPI networks as metric spaces and applied mathematical and computational algorithms to analyze their structural and functional properties. Our findings reveal that these networks share a conserved architecture across different cancer types, with central zones enriched in essential proteins and critical regulatory pathways. Notably, zones 1 and 2 of the cancer PPI networks are uniquely enriched in specific pathways, underscoring their importance in the progression of cancer. These results highlight the potential of metric-based analysis of PPI networks to uncover key molecular targets and accelerate drug discovery in oncology.

Keywords

Metric spaces; network modeling; cancer protein networks; core-periphery structure; drug discovery

1. Introduction

Cancer remains one of the leading causes of morbidity and mortality worldwide, representing a significant public health concern. In recent years, advances in high-throughput technologies (including genomics, transcriptomics, and proteomics) have enabled researchers to uncover shared molecular features across different cancer types. Such integrative analyses have revealed that cancers, despite their tissue-specific distinctions, often share dysregulated pathways involving oncogenes, tumor suppressors, and cell signaling networks [1,2,3,4,5]. These insights have motivated efforts to identify common therapeutic targets, particularly by focusing on proteins and pathways that intersect across multiple cancer types [6].

Among the most promising tools for capturing these shared molecular landscapes are PPI networks, which offer a systems-level view of cellular organization and are essential for nearly all biological processes, including those underlying cancer initiation, progression, and metastasis [7,8,9]. Mapping cancer-related PPI networks allows for the identification of key driver genes and interaction hubs that are frequently altered across malignancies, offering valuable targets for therapeutic intervention [10,11,12,13]. Structural analysis of these networks has become a powerful strategy in drug discovery, as it uncovers conserved interaction patterns and topological similarities across cancer types [14,15,16].

Recent studies have approached cancer PPI networks as metric spaces, using mathematical models to analyze their modular structure and organizational principles [17,18,19]. Centrality measures, such as closeness and eccentricity, have revealed hierarchical zones within the network, distinguishing central hubs involved in information processing and cell cycle regulation from peripheral nodes associated with metabolic functions [20,21,22]. Notably, enrichment analyses have shown that proteins located in the central zones are more likely to be essential and functionally specialized, highlighting their potential as drug targets [23,24].

Moreover, by integrating integrative cancer biology with network-based approaches, researchers have begun to construct integrated cancer PPI networks that highlight proteins shared among multiple cancer types. This framework not only identifies enriched biological pathways (such as oxidative phosphorylation, DNA repair, and immune regulation) but also reveals zones of functional convergence that could serve as universal therapeutic targets [25,26,27]. Several studies have demonstrated that different cancer types exhibit similarities in their PPI network topology, suggesting that a shared molecular infrastructure may support cancer pathogenesis across tissues [28,29,30].

Recent advances in graph learning and topological analysis of biological networks have introduced methods such as attributed graph embedding for protein-protein interaction link prediction [31], spectral clustering with graph structure learning [32], and affinity matrix construction integrating node attributes [33]. These methods excel in uncovering latent relationships and modular structures, but typically do not focus on stratifying networks into concentric layers around a topological center. Our metric-space zoning framework complements these approaches by explicitly identifying and functionally profiling core-periphery organization across 13 cancer types, enabling cross-cancer comparison of conserved functional hubs and facilitating integration with essential gene and drug-target datasets. This combination of topological zoning and comparative enrichment analysis provides biologically interpretable insights that extend beyond conventional clustering or embedding methods.

In this study, we investigated the structural similarities and differences among PPI networks across 13 types of cancer. While each cancer type has distinct molecular and clinical features, many share disrupted biological processes, including cell signaling, metabolism, and DNA repair. By constructing cancer PPI networks that integrate proteins and pathways shared across cancers, we aimed to uncover central zones enriched with essential proteins and functional pathways that may serve as universal therapeutic targets.

2. Materials and Methods

2.1 Data Sources

We considered the HFPIN [34], which contains 9,448 nodes and 181,706 interactions. Gene expression presence/absence calls were obtained for multiple cancer types, including breast, lung, kidney, pancreatic, liver, cervical, ovarian, glioblastoma, pituitary gland, glioma, fallopian tube, endometrium, and rectum. These data were retrieved from the Gene Expression Barcode Database (https://www.hsls.pitt.edu/obrc/index.php?page=URL20110523150503). Genes expressed in at least 99% of samples of a specific cancer were selected against the human hgu133 background [35]. The "consistently expressed" designation required presence calls in ≥99% of samples, ensuring robust inclusion across patients. Expression values were background-corrected to the hgu133 platform to maintain cross-cancer comparability.

2.2 Network Construction

We constructed cancer-specific protein interaction networks by mapping tumor-expressed proteins onto the human interactome (Figure 1). The algorithm followed these steps:

(1) Build a graph G across the binary protein–protein interaction list of HFPIN;

(2) Identify tumor proteins consistently expressed in cancers into a list L;

(3) Construct the interaction data between them from graph G if the proteins are in list L;

(4) Output the induced graph G'.

Click to view original image

Figure 1 Generation of cancer-specific protein interaction graphs from the human interactome.

2.3 Metric Space and Zoning Framework

To evaluate interaction networks as metric spaces, we applied a graph-theoretic approach in which cancer PPI networks were modeled by defining the topological distance between proteins as the shortest path between nodes. This approach formalizes the concept of a metric space, where distance reflects the number of steps (edges) separating two proteins. Using a Python wrapper around the C++ BOOST graph library (http://www.boost.org/), we employed Dijkstra's algorithm to compute shortest distances between all pairs of nodes. Based on these pairwise distances, the eccentricity of each node (defined as the maximum shortest path length from that node to any other node) was calculated. The node (or nodes) with the minimum eccentricity was designated as the network centre, representing the most centrally located point(s). Each remaining node was assigned to a "zone" based on its shortest distance from the centre. For instance, proteins located one step away formed zone 1, two steps away formed zone 2, and so forth (Figure 2). Zone assignments were validated by cross-referencing with alternative centrality measures to ensure consistency. This zoning framework enabled structural and functional analysis, revealing patterns such as central enrichment of essential or functionally specialized proteins.

Click to view original image

Figure 2 Cancer PPI networks organized into zones according to their distance from the network centre.

In contrast to earlier applications of zoning to single network types, here the method is extended to a multi-cancer setting, coupled with zone-specific enrichment analysis against curated biological databases. This ensures that the observed core enrichment patterns are both reproducible and biologically meaningful.

2.4 Functional Enrichment Analysis

To assess the robustness of our zoning framework, we performed two additional analyses: (i) enrichment significance testing using FDR correction (q < 0.05), and (ii) leave-one-cancer-type-out cross-validation to evaluate the stability of identified central hubs and enriched pathways. Across all tests, the zoning structure, core hubs, and zone 1-2 enrichment profiles remained stable, providing support for the reliability of our approach.

For enrichment analysis, we used the Comparative Toxicogenomics Database (CTD; https://ctdbase.org/). A p-value threshold of 0.01 was chosen to define significance. Results included enriched gene sets, statistical significance, and enrichment scores, providing insight into the biological processes and molecular functions associated with the input gene list.

2.5 Protein Class Annotation

We evaluated the distribution of oncogenes, tumor suppressors, and their associated pathways using genome-wide sequencing studies of cancer [36]. To identify key proteins within the cancer PPI networks, we intersected a list of previously defined essential proteins (derived from knockout phenotypes of mouse orthologs) with the proteins assigned to each zone [37]. This allowed us to assess the enrichment of essential proteins across zones and highlight the functional significance of core zones.

2.6 Robustness and Random Graph Controls

To test the biological significance of the observed topologies, we compared each cancer PPI network to a large set of computationally generated uniform random power-law graphs with similar node and edge counts [38]. While all cancer PPI networks exhibited a single centre and large diameter, the random equivalents often contained multiple centres and smaller diameters. Cancer PPI networks also had a higher proportion of low-degree "quill" nodes and a few extremely high-degree hubs, absent in random graphs. Moreover, the cancer networks consistently formed a single giant component, in contrast to the fragmented structures of many random equivalents. These differences indicate that the observed zoning patterns are not random artifacts but rather reflect a biologically evolved organization.

2.7 Visualization

Zone-level enrichment heatmaps were generated in Python (matplotlib), with pathways as columns and zones as rows. Cells display average (%) enrichment; missing cells indicate pathways not observed at FDR-controlled significance in that zone.

3. Results

3.1 Infrastructure of Individual Proteins in Cancer PPI Networks

We considered tumor proteins in HFPIN that are consistently expressed. Most cancers exhibited a high number of consistently expressed proteins (namely, rectum, endometrium, fallopian, glioma, pituitary, glioblastoma, ovary, cervix, liver, and pancreas), whereas breast, lung, and kidney cancers had the fewest. Across cancers, ~600 proteins were consistently expressed on average. In addition, tumor PPI networks contained fewer expressed proteins than their corresponding normal tissues (Table 1).

Table 1 Distribution of the normal and tumor proteins in HFPIN.

3.2 Similarity between Cancer PPI Networks as Metric Spaces

We constructed 13 protein interaction networks representing pancreatic, kidney, lung, breast, cervical, glioblastoma, liver, ovarian, fallopian, glioma, rectal, endometrial, and pituitary cancers. Network properties and their structural similarities are summarized in Table 2 and Table 3. Several recurring features emerged:

(i) Ten PPI-nets (breast, kidney, cervix, pancreas, ovary, glioblastoma, pituitary, glioma, liver, and lung) share RPS27A as the centre. RPS27A plays a significant role in targeting cellular proteins [39], underscoring its potential as a cancer drug target, particularly given its central role in protein synthesis. Two cancers (endometrium and fallopian) have XPO1 as the centre; XPO1 modulates the localization of cyclin B, MPAK, and MAPKAP kinase 2 [40]. Finally, rectum cancer has TP53 as its centre; TP53 is a core cell-cycle/apoptosis and DNA-damage control pathway component and a tumor suppressor in many tumor types [41].

(ii) Proteins closer to the topological centre tend to have a higher degree.

(iii)The majority of nodes reside within zones 1-3 relative to the centre.

(iv)Nodes farther from the centre predominantly have low degree; at the periphery, most are single-degree ("quill") proteins.

(v) Most PPI-nets contain six zones around the centre; the fallopian, lung, and liver have five. Consistent with [18], modelling these networks with distance-respecting topology reveals a densely connected core that thins towards the periphery, ending in quills (degree = 1). Zone 1 is highly connected and contains very few low-degree proteins. The proportion of quills increases approximately exponentially with distance from the centre.

Table 2 Similarity between cancer PPI networks.

Table 3 Distribution and similarity of zones around the cancer PPI networks.

Robustness analyses confirmed the stability of these findings, as leave-one-cancer-type-out analysis preserved >85% of enriched pathways in zones 1-3 across cancers. All reported pathway enrichments are FDR-adjusted (q < 0.05) (Table S1).

3.3 Similarity Based on Dominant Biological Functions

Enrichment analyses revealed similar dominant biological functions across cancers. The average percentages were 31%, 24.9%, 19.7%, 18.6%, 18.3%, 13.1%, 12.6%, 9.9%, 7.7%, and 6% for gene expression, disease, metabolism of RNA, metabolism of proteins, immune system, signal transduction, 3'-UTR-mediated translational regulation, cell cycle, mitotic M-M/G1 phases, and apoptosis, respectively (Tables S2-S14). Further, their Representation of immune system pathways appeared approximately equal across cancers, consistent with the immune hallmarks of cancer [26,42]. Moreover, immune surveillance and recognition/clearance of nascent tumors have long been described as central immune functions [27,28].

We next examine cancer-specific PPI-nets to identify pathways that are uniquely enriched across zones of each cancer type. In general, central zones are implicated in sensing/regulatory functions that control cancer, whereas peripheral zones diversify into routine metabolic functions.

3.4 Zone-Level Biological Observations Across Cancers

3.4.1 Cancers with RPS27A as Centre

We refer to cancers centred on RPS27A as "RPS27A cancers". In 10 such cancers (breast, kidney, cervix, pancreas, ovary, glioblastoma, pituitary, glioma, liver, and lung), zone 1 exhibits the highest percentage of proteins involved in critical functional pathways, followed by zone 2; peripheral zones are enriched for metabolic pathways (Tables S15-S27). Three pathways are uniquely enriched in RPS27A cancers: in zone 1, 3'-UTR-mediated translational regulation and ribosome; in zone 2, mRNA processing. Figure 3, Figure 4, Figure 5 present zone-specific enrichment heatmaps.

Click to view original image

Figure 3 RPS27A-centred cancers: Zone-specific enrichment across major functional pathways.

Click to view original image

Figure 4 XPO1-centred cancers: Zone-specific enrichment.

Click to view original image

Figure 5 TP53-centred cancers: zone-specific enrichment.

Zone 1. The summary of functional pathways in zone 1 of Table 4 summarizes zone 1 pathways for RPS27A cancers. The average percentages were 86%, 83.7%, 83.5%, 64.8%, 64.8%, and 63% for gene expression, RNA metabolism, 3'-UTR-mediated translational regulation, and ribosome function, respectively. The latter two are uniquely enriched in zone 1. 3'-UTR-mediated translational regulation is implicated in tumorigenesis and metastasis [29], and ribosome biogenesis has been linked to malignant progression and is inhibited by several anti-cancer drugs [30,41]. These observations underscore the biological importance of zone 1 proteins and their potential as novel drug targets.

Table 4 Summary of functional pathways in zone 1 of RPS27A cancers.

Breast cancer differs from other RPS27A cancers in lacking signal transduction enrichment in zone 1. Nevertheless, given the strong enrichment of this pathway in the different cancers, zone 1 proteins from this functional class may still represent attractive targets.

Zone 2. Table 5 summarizes zone-2 pathways for RPS27A cancers. Gene expression (29.1%), the immune system (21%), and mRNA processing (20.8%) are the primary areas of focus. As noted earlier, mRNA processing is uniquely enriched in zone 2 of these cancers and is clinically relevant as both a driver of disease states and a potential biomarker. Additionally, mRNA processing targets and ribosomes are absent in zone 2 [42,43,44]. Signal transduction and cell cycle do not appear in breast cancer in this zone, and mitotic M-M/G1 phases are absent in breast, kidney, liver, and lung. Overall, zone 2 shows a less uniform functional similarity than zone 1.

Table 5 Summary of functional pathways in zone 2 of RPS27A cancers.

Zone 3. Table 6 lists zone-3 pathways for RPS27A cancers. As expected, average percentages of significant pathways are lower than in zones 1-2. Functions dominant in zones 1-2 largely disappear, except the immune system (16.9%), gene expression (12.9%), and metabolism of proteins (11.5%). This pattern is consistent with the hypothesis that functional dominance decreases with distance from the centre.

Table 6 Summary of functional pathways in zone 3 of RPS27A cancers.

3.4.2 Cancers with XPO1 as Centre

Cancers centred on XPO1 are referred to as "XPO1 cancers." Overall enrichment dominance is lower than in RPS27A cancers. Pathways uniquely enriched in zone 1 of RPS27A cancers become uniquely enriched in zone 2 of XPO1 cancers; mRNA processing appears in both zones but not in zone 3.

Zone 1. Table 7 summarizes zone 1 pathways in XPO1 cancers. The most prominent pathways are disease (34%), cell cycle (32.9%), immune system (32.5%), mitotic M-M/G1 (24.7%), and gene expression (20.5%). Notably, signal transduction, protein metabolism, 3'-UTR-mediated translational regulation, and ribosomes are not represented here, despite being dominant in zone 1 of RPS27A cancers. Mitotic anti-cancer strategies have been revisited recently [45].

Table 7 Summary of functional pathways in zone 1 of XPO1 cancers.

Zone 2. Table 8 summarizes zone-2 pathways in XPO1 cancers. Gene expression increases to 40.7% (approximately double that of zone 1). Pathways uniquely enriched in zone 1 of RPS27A cancers (e.g., 3'-UTR-mediated translational regulation, ribosome) become uniquely enriched in zone 2 here. Signal transduction also appears in zone 2. These observations reinforce the potential of zone-2 proteins in XPO1 cancers as novel targets.

Table 8 Summary of functional pathways in zone 2 of XPO1 cancers.

Zone 3. Again, in zone 3, several functions are absent (disease, 3'-UTR-mediated translational regulation, ribosome, and mRNA processing). Signal transduction appears in the endometrium but not the fallopian tubes. Overall averages in zone 3 are weaker than those in zones 1-2 (Table 9).

Table 9 Summary of functional pathways in zone 3 of XPO1 cancers.

3.4.3 Cancers with TP53 as Centre

Table 10 presents functional pathways by zone for TP53-centred cancers. Zone 1 shows unique enrichment of the p53 signaling pathway; p53 is engaged in stress responses and supports targeted therapy development [46,47]. In zone 2, 3'-UTR-mediated translational regulation, ribosome, and mRNA processing are uniquely enriched (again highlighting the central role of zones 1-2 in cancer biology and drug targeting).

Table 10 Functional pathways in TP53 cancer zones.

3.5 Distribution of Essential, Signaling, Growth, Cell Cycle, Tumor Suppressor, Oncogenic, and Therapeutic-Target Classes

Class-distribution patterns were broadly similar across cancers (Table 11). Despite zone-specific differences in protein counts, percentage distributions across zones were comparable.

Table 11 Distribution of essential, signaling, growth, cell cycle, tumor suppressor, oncogenic, and therapeutic targets in cancer PPI-nets based on distance from the topological centre.

Signaling roles were the most prominent (34.8%), followed by essential (22.2%) and cell cycle (11.3%) functions. Below, we detail how each class is distributed across zones for each cancer type.

3.6 Protein-Class Distributions Across Zones

Using a conserved human essential gene list (Mouse Genome Database) [37], we observed the highest percentages of crucial proteins in central zones: zones 1-5 comprised 10.9%, 40%, 33%, 9.2%, and 1.1%, respectively (Table S28).

A similar pattern held for signaling proteins: 10%, 43%, 31.1%, 8.4%, and 1.0% in zones 1-5, respectively (Table S29).

Growth proteins (though fewer) were also central, with percentages of 12.7%, 34.5%, 43.5%, 8.9%, and 0.3% across zones 1-5 (Table S30), supporting the view that central zones are promising for drug targeting.

For the cell cycle, zones 1-2 dominated at 40.8% and 40.6%, followed by 15.5%, 1.7%, and 0%, respectively (Table S31). Dysregulated cell-cycle control underlies uncontrolled growth and genomic damage (hallmarks of cancer [48]).

Drug targets concentrated in zones 1-3: 8.7%, 45.7%, 39%, 5%, 0% (Table S32), further underscoring the translational value of central zones.

Oncogenes peaked in zone 2 (Table S33): 1.8%, 85%, 10%, 1.9%, and 0.9% across zones 1-5. Prior work links oncogenes/tumor suppressors to metastatic progression and essential tumor functions [47,49,50], supporting the focus on zone 2 as a reservoir of therapeutically relevant proteins.

Finally, tumor suppressors were enriched in zones 1-3, with high representation 1-2: 30.3%, 37.6%, 25.3%, 5.6%, and 0.9% across zones 1-5 (Table S34). Together with the oncogene pattern, this supports prioritizing central zones for target discovery.

4. Discussion

We introduced a network-biological approach that models cancer PPI networks as metric spaces. This framework reveals a conserved core-periphery organization across cancers, with central proteins (e.g., RPS27A, XPO1, and TP53) occupying the most connected zones and participating in key cellular programs. For instance, RPS27A is implicated in ribosome biogenesis and protein degradation [51], XPO1 regulates nuclear export and contributes to drug resistance [52], and TP53 is a critical tumor suppressor governing cell-cycle checkpoints and apoptosis [53].

Our work complements graph-learning methods by providing a mathematically transparent, interpretable scheme for zone identification. The observed centrality of RPS27A, XPO1, and TP53 is consistent with graph-embedding findings [54]. We extend these observations by demonstrating zone-specific functional enrichment across various types of cancer. Thus, integrating metric-based zoning with functional annotation offers a roadmap to prioritize targets supported by both topology and biology.

Regarding zone-level enrichment, cancers centred on RPS27A show clear dominance of core pathways in central zones. Zone 1 is the most dominant, uniquely enriched for 3'-UTR-mediated translational regulation and ribosome biogenesis, both of which are implicated in oncogenesis and are therapeutically actionable [55,56]. Zone 2 remains functionally important and is uniquely enriched for mRNA processing, whereas zone 3 is comparatively attenuated. These patterns highlight the critical role of near-centre proteins as potential drug targets, while more peripheral zones tend to be enriched for metabolic functions. Notably, signal-transduction pathways were not detected in zone 1 of RPS27A cancers (Figure 6).

Click to view original image

Figure 6 Dominant zones in GO enrichment functions in RPS27A cancers.

In XPO1-centred cancers, overall enrichment is slightly less pronounced than in RPS27A cancers, yet key shifts are evident. Pathways uniquely enriched in zone 1 of RPS27A cancers are also present in zone 2 of XPO1 cancers. Conversely, zone 1 of XPO1 cancers is enriched in cell-cycle, immune, and mitotic M-M/G1 processes, reinforcing the centrality of zone 1 proteins to cancer biology and the discovery of potential targets. Zone 2 exhibits dominance in RNA metabolism, gene expression, and protein metabolism, whereas zone 3 mirrors RPS27A cancers with predominantly metabolic activity (Figure 7).

Click to view original image

Figure 7 Dominant zones in GO enrichment functions in XPO1 cancers.

For TP53-centred cancers (notably rectum), overall enrichment is lower than in RPS27A and XPO1 cancers; however, zone 2 shows unique enrichment for 3'-UTR-mediated translational regulation, ribosome biogenesis, and mRNA processing, further underscoring the functional relevance of central zones (Figure 8) and aligning with their roles in cancer progression and regulation [57,58].

Click to view original image

Figure 8 Dominant zones in GO enrichment functions in TP53 cancers.

A significant finding is the tendency of essential genes, oncogenes, and tumor suppressors to cluster in core zones (zones 1-3). For instance, 22.2% of essential genes and 34.8% of signaling proteins are central, consistent with theories of network centrality and tumor survival [59,60]. Moreover, evidence from the cancer dependency map indicates that centrally located proteins are more likely to be critical dependencies [61].

Several central proteins highlighted here already have experimental or clinical validation. Inhibiting ribosomal RNA synthesis (e.g., CX-5461) demonstrates antitumor activity, aligning with the role of RPS27A in ribosome biogenesis/protein turnover [62,63]. XPO1 is clinically targetable with selinexor, which disrupts nuclear export and impairs tumor proliferation [64]. TP53 remains the prototypical tumor suppressor with ongoing efforts to restore function clinically. In addition, enrichment of 3'-UTR regulation, ribosome biogenesis, and mRNA processing is supported by experimental studies linking the disruption of these processes to oncogenesis [65]. Collectively, these observations externally validate our computational predictions and strengthen their translational relevance, while also motivating laboratory experiments to confirm novel predictions.

Although based on computational analyses, our results emphasize the need for experimental validation. Future studies should test whether targeting central-zone proteins improves therapeutic outcomes and examine how peripheral, metabolically enriched proteins may contribute to drug resistance.

In summary, modeling cancer PPI networks as metric spaces provides a powerful lens for uncovering shared network architectures and functional liabilities across cancers. The repeated dominance of ribosome, mRNA processing, and 3'-UTR-mediated regulation in zones 1-2 indicates that these zones house key, biologically central proteins with therapeutic promise [66,67]. A data-level limitation is the absence of patient and subtype-specific variation in the expression inputs. While our consensus-network strategy facilitates cross-cancer comparability, it may overlook clinically meaningful heterogeneity. Future work should incorporate individual-level datasets to delineate pan-cancer targets alongside subtype-restricted vulnerabilities.

A methodological limitation is that eccentricity-based zoning assumes an essentially connected network with a single dominant centre (or a small co-central set). This assumption held across the 13 cancers analysed, but networks with unusually high modularity, multiple dense hubs, or fragmented components may deviate. In such cases, zones should be computed per significant component or module to preserve biological interpretability. Extending the zoning framework to multi-centre or fragmented architectures is an essential direction for methodological development.

5. Conclusions

The construction of cancer PPI networks provides a valuable framework for drug discovery by offering a systems-level view of the molecular interactions underlying cancer progression. Identifying dysregulated driver proteins and pathways within these networks can guide the development of therapeutic strategies. In this study, we modeled cancer PPI networks as metric spaces and applied mathematical and computational algorithms to uncover critical structural and functional properties.

Our findings demonstrate that cancer PPI networks share a conserved metric-space organization, with central zones enriched for essential proteins and key functional pathways. These proteins control fundamental cellular decisions, including pathway regulation, and are disproportionately concentrated in the core regions of the cell. Notably, zones 1 and 2 exhibited unique enrichment in pathways central to cancer progression, underscoring their translational significance.

In conclusion, metric-space analysis of cancer PPI networks represents a powerful and innovative approach for rational drug discovery. By highlighting conserved network structures and central functional hubs, this framework facilitates the identification of promising therapeutic targets across diverse cancer types.

Acknowledgments

This work is based on research supported by the Faculty of Natural Science Research Office at University of the Western Cape. The opinions and conclusions expressed are those of the author and should not necessarily be attributed to the research office.

Author Contributions

The author solely contributed to all aspects of the study, including conceptualization, data curation, formal analysis, investigation, methodology, software, writing – original draft, and writing – review & editing.

Competing Interests

The authors declare that there is no competing interest exist.

AI-Assisted Technologies Statement

I confirm that I used ChatGPT only for the purpose of basic grammar correction and language enhancement during the preparation of this manuscript. All AI-assisted text has been carefully reviewed and edited by me for accuracy. I take full responsibility for the content of the manuscript.

Additional Materials

The following additional materials are uploaded at the page of this paper.

  1. Table S1: FDR corrected enrichment analysis.
  2. Table S2: Top pathway functions of breast cancer.
  3. Table S3: Top pathway functions of pituitary cancer.
  4. Table S4: Top pathway functions of lung cancer.
  5. Table S5: Top pathway functions of kidney cancer.
  6. Table S6: Top pathway functions of liver cancer.
  7. Table S7: Top pathway functions of cervix cancer.
  8. Table S8: Top pathway functions of ovary cancer.
  9. Table S9: Top pathway functions of fallopian cancer.
  10. Table S10: Top pathway functions of endometrium cancer.
  11. Table S11: Top pathway functions of pancreas cancer.
  12. Table S12: Top pathway functions of glioblastoma cancer.
  13. Table S13: Top pathway functions of glioma cancer.
  14. Table S14: Top pathway functions of rectum cancer.
  15. Table S15: Summary of functional pathways in zones of breast cancer.
  16. Table S16: Summary of functional pathways in zones of pituitary cancer.
  17. Table S17: Summary of functional pathways in zones of kidney cancer.
  18. Table S18: Summary of functional pathways in zones of liver cancer.
  19. Table S19: Summary of functional pathways in zones of pancreas cancer.
  20. Table S20: Summary of functional pathways in zones of cervix cancer.
  21. Table S21: Summary of functional pathways in zones of ovary cancer.
  22. Table S22: Summary of functional pathways in zones of glioblastoma cancer.
  23. Table S23: Summary of functional pathways in zones of glioma cancer.
  24. Table S24: Summary of functional pathways in zones of fallopian cancer.
  25. Table S25: Summary of functional pathways in zones of fallopian cancer.
  26. Table S26: Summary of functional pathways in zones of endometrium cancer.
  27. Table S27: Summary of functional pathways in zones of recuum cancer.
  28. Table S28: Distribution of essential proteins in zones of cancer PINs.
  29. Table S29: Distribution of signalling proteins in zones of cancer PINs.
  30. Table S30: Distribution of growth proteins in zones of cancer PINs.
  31. Table S31: Distribution of cell cycle proteins in zones of cancer PINs.
  32. Table S32: Distribution of drug target proteins in zones of cancer PINs.
  33. Table S33: Distribution of oncogene proteins in zones of cancer PINs.
  34. Table S34: Distribution of suppressors proteins in zones of cancer PINs.

References

  1. Hoadley KA, Yau C, Hinoue T, Wolf DM, Lazar AJ, Drill E, et al. Cell-of-origin patterns dominate the molecular classification of 10,000 tumors from 33 types of cancer. Cell. 2018; 173: 291-304. [CrossRef] [Google scholar] [PubMed]
  2. Sanchez-Vega F, Mina M, Armenia J, Chatila WK, Luna A, La KC, et al. Oncogenic signaling pathways in the cancer genome atlas. Cell. 2018; 173: 321-337. [CrossRef] [Google scholar] [PubMed]
  3. Weinstein JN, Collisson EA, Mills GB, Shaw KR, Ozenberger BA, Ellrott K, et al. The cancer genome atlas pan-cancer analysis project. Nat Genet. 2013; 45: 1113-1120. [CrossRef] [Google scholar] [PubMed]
  4. Hoadley KA, Yau C, Wolf DM, Cherniack AD, Tamborero D, Ng S, et al. Multiplatform analysis of 12 cancer types reveals molecular classification within and across tissues of origin. Cell. 2014; 158: 929-944. [CrossRef] [Google scholar] [PubMed]
  5. Panneerselvam J, Mohandoss P, Patel R, Gillan H, Li M, Kumar K, et al. DCLK1 regulates tumor stemness and cisplatin resistance in non-small cell lung cancer via ABCD-member-4. Mol Ther Oncol. 2020; 18: 24-36. [CrossRef] [Google scholar] [PubMed]
  6. Srihari S, Madhamshettiwar PB, Song S, Liu C, Simpson PT, Khanna KK, et al. Complex-based analysis of dysregulated cellular processes in cancer. BMC Syst Biol. 2014; 8: S1. [CrossRef] [Google scholar] [PubMed]
  7. Fonseca A, Gubitoso MD, Reis MS, De Souza SJ, Barrera J. A new approach for identification of cancer-related pathways using protein networks and genomic data. Cancer Inform. 2015; 14: CIN-S30800. [CrossRef] [Google scholar] [PubMed]
  8. Kanehisa M, Furumichi M, Tanabe M, Sato Y, Morishima K. KEGG: New perspectives on genomes, pathways, diseases and drugs. Nucleic Acids Res. 2017; 45: D353-D361. [CrossRef] [Google scholar] [PubMed]
  9. Goh KI, Cusick ME, Valle D, Childs B, Vidal M, Barabási AL. The human disease network. Proc Natl Acad Sci USA. 2007; 104: 8685-8690. [CrossRef] [Google scholar] [PubMed]
  10. Jonsson PF, Bates PA. Global topological features of cancer proteins in the human interactome. Bioinformatics. 2006; 22: 2291-2297. [CrossRef] [Google scholar] [PubMed]
  11. Wachi S, Yoneda K, Wu R. Interactome-transcriptome analysis reveals the high centrality of genes differentially expressed in lung cancer tissues. Bioinformatics. 2005; 21: 4205-4208. [CrossRef] [Google scholar] [PubMed]
  12. Aragues R, Sander C, Oliva B. Predicting cancer involvement of genes from heterogeneous data. BMC Bioinform. 2008; 9: 172. [CrossRef] [Google scholar] [PubMed]
  13. Jonsson PF, Cavanna T, Zicha D, Bates PA. Cluster analysis of networks generated through homology: Automatic identification of important protein communities involved in cancer metastasis. BMC Bioinform. 2006; 7: 2. [CrossRef] [Google scholar] [PubMed]
  14. Zhang W, Wang SL. A novel method for identifying the potential cancer driver genes based on molecular data integration. Biochem Genet. 2020; 58: 16-39. [CrossRef] [Google scholar] [PubMed]
  15. Bandyopadhyay S, Chiang CY, Srivastava J, Gersten M, White S, Bell R, et al. A human MAP kinase interactome. Nat Methods. 2010; 7: 801-805. [CrossRef] [Google scholar] [PubMed]
  16. Yilmaz A, Peretz M, Aharony A, Sagi I, Benvenisty N. Defining essential genes for human pluripotent stem cells by CRISPR–Cas9 screening in haploid cells. Nat Cell Biol. 2018; 20: 610-619. [CrossRef] [Google scholar] [PubMed]
  17. Vinogradov AE. Modularity of cellular networks shows general center-periphery polarization. Bioinformatics. 2008; 24: 2814-2817. [CrossRef] [Google scholar] [PubMed]
  18. Fadhal E, Gamieldien J, Mwambene EC. Protein interaction networks as metric spaces: A novel perspective on distribution of hubs. BMC Syst Biol. 2014; 8: 6. [CrossRef] [Google scholar] [PubMed]
  19. Fadhal E, Mwambene EC, Gamieldien J. Modelling human protein interaction networks as metric spaces has potential in disease research and drug target discovery. BMC Syst Biol. 2014; 8: 68. [CrossRef] [Google scholar] [PubMed]
  20. Fadhal E, Gamieldien J, Mwambene EC. Self-similarity of human protein interaction networks: A novel strategy of distinguishing proteins. Sci Rep. 2015; 5: 7628. [CrossRef] [Google scholar] [PubMed]
  21. Johnson Z, Anderson D, Cheung MS, Bohutskyi P. Gene network centrality analysis identifies key regulators coordinating day-night metabolic transitions in Synechococcus elongatus PCC 7942 despite limited accuracy in predicting direct regulator-gene interactions. Front Microbiol. 2025; 16: 1569559. [CrossRef] [Google scholar] [PubMed]
  22. Fadhal E. A comprehensive analysis of the PI3K/AKT pathway: Unveiling key proteins and therapeutic targets for cancer treatment. Cancer Inform. 2023; 22. doi: 10.1177/11769351231194273. [CrossRef] [Google scholar] [PubMed]
  23. Alqattan HR, Fadhal E. Exploring the significance of zone 1 in human protein interaction networks: Identifying potential therapeutic targets for cancer treatment. Oncol Radiother. 2023; 17: 249-253. [Google scholar]
  24. Fadhal E. Exploring lung cancer protein network: Understanding structure and function through metric space modeling. Eur J Pure Appl Math. 2024; 17: 905-921. [CrossRef] [Google scholar]
  25. Groarke JM, O'Reilly S, Burke SM, Sweeney KJ, Newell J, Flanagan M, et al. Characteristics and outcomes of breast cancer in Black patients in a majority White population. Breast Cancer Res Treat. 2021; 186: 721-728. [Google scholar]
  26. Lawrence HS. Delayed sensitivity and homograft sensitivity. Annu Rev Med. 1960; 11: 207-230. [CrossRef] [Google scholar] [PubMed]
  27. Burnet FM. The concept of immunological surveillance. In: Immunological Aspects of Neoplasia. S.Karger AG; 1970. doi: 10.1159/000386035. [Google scholar]
  28. Chaudhury A, Hussey GS, Howe PH. 3'-UTR-mediated post-transcriptional regulation of cancer metastasis: Beginning at the end. RNA Biol. 2011; 8: 595-599. [CrossRef] [Google scholar] [PubMed]
  29. Ruggero D, Pandolfi PP. Does the ribosome translate cancer? Nat Rev Cancer. 2003; 3: 179-192. [CrossRef] [Google scholar] [PubMed]
  30. Burger K, Mühl B, Harasim T, Rohrmoser M, Malamoussi A, Orban M, et al. Chemotherapeutic drugs inhibit ribosome biogenesis at various levels. J Biol Chem. 2010; 285: 12416-12425. [CrossRef] [Google scholar] [PubMed]
  31. Nasiri E, Berahmand K, Rostami M, Dabiri M. A novel link prediction algorithm for protein-protein interaction networks by attributed graph embedding. Comput Biol Med. 2021; 137: 104772. [CrossRef] [Google scholar] [PubMed]
  32. Berahmand K, Saberi-Movahed F, Sheikhpour R, Li Y, Jalili M. A comprehensive survey on spectral clustering with graph structure learning. arXiv. 2025. doi: 10.48550/arXiv.2501.13597. [Google scholar]
  33. Berahmand K, Nasiri E, Li Y. Spectral clustering on protein-protein interaction networks via constructing affinity matrix using attributed graph embedding. Comput Biol Med. 2021; 138: 104933. [CrossRef] [Google scholar] [PubMed]
  34. Wu G, Feng X, Stein L. A human functional protein interaction network and its application to cancer data analysis. Genome Biol. 2010; 11: R53. [CrossRef] [Google scholar] [PubMed]
  35. McCall MN, Uppal K, Jaffee HA, Zilliox MJ, Irizarry RA. The Gene Expression Barcode: Leveraging public data repositories to begin cataloging the human and murine transcriptomes. Nucleic Acids Res. 2011; 39: D1011-D1015. [CrossRef] [Google scholar] [PubMed]
  36. Vogelstein B, Papadopoulos N, Velculescu VE, Zhou S, Diaz Jr LA, Kinzler KW. Cancer genome landscapes. Science. 2013; 339: 1546-1558. [CrossRef] [Google scholar] [PubMed]
  37. Bult CJ, Eppig JT, Blake JA, Kadin JA, Richardson JE, Mouse Genome Database Group. The mouse genome database: Genotypes, phenotypes, and models of human disease. Nucleic Acids Res. 2012; 41: D885-D891. [CrossRef] [Google scholar] [PubMed]
  38. Aiello W, Chung F, Lu L. A random graph model for power law graphs. Exp Math. 2001; 10: 53-66. [CrossRef] [Google scholar]
  39. Benson D, Boguski M, Lipman DJ, Ostell J. The national center for biotechnology information. Genomics. 1990; 6: 389-391. [CrossRef] [Google scholar] [PubMed]
  40. Cooper DN, Ball EV, Krawczak M. The human gene mutation database. Nucleic Acids Res. 1998; 26: 285-287. [CrossRef] [Google scholar] [PubMed]
  41. Danckwardt S, Hentze MW, Kulozik AE. 3' end mRNA processing: Molecular mechanisms and implications for health and disease. EMBO J. 2008; 27: 482-498. [CrossRef] [Google scholar] [PubMed]
  42. Singh P, Alley TL, Wright SM, Kamdar S, Schott W, Wilpan RY, et al. Global changes in processing of mRNA 3' untranslated regions characterize clinically distinct cancer subtypes. Cancer Res. 2009; 69: 9422-9430. [CrossRef] [Google scholar] [PubMed]
  43. Venables JP, Koh CS, Froehlich U, Lapointe E, Couture S, Inkel L, et al. Multiple and specific mRNA processing targets for the major human hnRNP proteins. Mol Cell Biol. 2008; 28: 6033-6043. [CrossRef] [Google scholar] [PubMed]
  44. Chan KS, Koh CG, Li HY. Mitosis-targeted anti-cancer therapies: Where they stand. Cell Death Dis. 2012; 3: e411. [CrossRef] [Google scholar] [PubMed]
  45. Bourdon JC. p53 and its isoforms in cancer. Br J Cancer. 2007; 97: 277-282. [CrossRef] [Google scholar] [PubMed]
  46. Vazquez A, Bond EE, Levine AJ, Bond GL. The genetics of the p53 pathway, apoptosis and cancer therapy. Nat Rev Drug Discov. 2008; 7: 979-987. [CrossRef] [Google scholar] [PubMed]
  47. Collins K, Jacks T, Pavletich NP. The cell cycle and cancer. Proc Natl Acad Sci USA. 1997; 94: 2776-2778. [CrossRef] [Google scholar] [PubMed]
  48. Min J, Zaslavsky A, Fedele G, McLaughlin SK, Reczek EE, De Raedt T, et al. An oncogene–tumor suppressor cascade drives metastatic prostate cancer by coordinately activating Ras and nuclear factor-κB. Nature medicine. 2010; 16: 286-294. [CrossRef] [Google scholar] [PubMed]
  49. Münger K. Disruption of oncogene/tumor suppressor networks during human carcinogenesis. Cancer Investig. 2002; 20: 71-81. [CrossRef] [Google scholar] [PubMed]
  50. Lee EY, Muller WJ. Oncogenes and tumor suppressor genes. Cold Spring Harb Perspect Biol. 2010; 2: a003236. [CrossRef] [Google scholar] [PubMed]
  51. Luo J, Zhao H, Chen L, Liu M. Multifaceted functions of RPS27a: An unconventional ribosomal protein. J Cell Physiol. 2023; 238: 485-497. [CrossRef] [Google scholar] [PubMed]
  52. Zeng M, Li M, Fei Z, Wu FX, Li Y, Pan Y, et al. A deep learning framework for identifying essential proteins by integrating multiple types of biological information. IEEE/ACM Trans Comput Biol Bioinform. 2019; 18: 296-305. [CrossRef] [Google scholar] [PubMed]
  53. Azmi AS, Uddin MH, Mohammad RM. The nuclear export protein XPO1—from biology to targeted therapy. Nature Rev Clin Oncol. 2021; 18: 152-169. [CrossRef] [Google scholar] [PubMed]
  54. Giono LE, Manfredi JJ. The p53 tumor suppressor participates in multiple cell cycle checkpoints. J Cell Physiol. 2006; 209: 13-20. [CrossRef] [Google scholar] [PubMed]
  55. Kang J, Brajanovski N, Chan KT, Xuan J, Pearson RB, Sanij E. Ribosomal proteins and human diseases: Molecular mechanisms and targeted therapy. Signal Transduct Target Ther. 2021; 6: 323. [CrossRef] [Google scholar] [PubMed]
  56. Mayr C. What are 3' UTRs doing? Cold Spring Harb Perspect Biol. 2019; 11: a034728. [CrossRef] [Google scholar] [PubMed]
  57. Kamath D, Iwakuma T, Bossmann SH. Therapeutic potential of combating cancer by restoring wild-type p53 through mRNA nanodelivery. Nanomed Nanotechnol Biol Med. 2024; 56: 102732. [CrossRef] [Google scholar] [PubMed]
  58. Gandhi UH, Senapedis W, Baloglu E, Unger TJ, Chari A, Vogl D, et al. Clinical implications of targeting XPO1-mediated nuclear export in multiple myeloma. Clin Lymphoma Myeloma Leuk. 2018; 18: 335-345. [CrossRef] [Google scholar] [PubMed]
  59. Rout T, Mohapatra A, Kar M, Muduly DK. Centrality-Based Approach for Identifying Essential Cancer Proteins in PPI Networks. SN Comput Sci. 2024; 6: 6. [CrossRef] [Google scholar]
  60. Patil SS, Roberts SA, Gebremedhin AH. Network analysis of driver genes in human cancers. Front Bioinform. 2024; 4: 1365200. [CrossRef] [Google scholar] [PubMed]
  61. Shimada K, Bachman JA, Muhlich JL, Mitchison TJ. shinyDepMap, a tool to identify targetable cancer genes and their functional connections from Cancer Dependency Map data. Elife. 2021; 10: e57116. [CrossRef] [Google scholar] [PubMed]
  62. Hongyan L, Hong Z, Guomin H, Zhitong B, Duling X, Jiadi L, et al. Loss of RPS27a expression regulates the cell cycle, apoptosis, and proliferation via the RPL11-MDM2-p53 pathway in lung adenocarcinoma cells. J Exp Clin Cancer Res. 2022; 41: 1-20. [CrossRef] [Google scholar] [PubMed]
  63. Elhamamsy AR, Metge BJ, Alsheikh HA, Shevde LA, Samant RS. Ribosome biogenesis: A central player in cancer metastasis and therapeutic resistance. Cancer Res. 2022; 82: 2344-2353. [CrossRef] [Google scholar] [PubMed]
  64. Azizian NG, Li Y. XPO1-dependent nuclear export as a target for cancer therapy. J Hematol Oncol. 2020; 13: 61. [CrossRef] [Google scholar] [PubMed]
  65. Belin S, Beghin A, Solano-Gonzàlez E, Bezin L, Brunet-Manquat S, Textoris J, et al. Dysregulation of ribosome biogenesis and translational capacity is associated with tumor progression of human breast cancer cells. PLoS One. 2009; 4: e7147. [CrossRef] [Google scholar] [PubMed]
  66. Pawson T, Linding R. Network medicine. FEBS Lett. 2008; 582: 1266-1270. [CrossRef] [Google scholar] [PubMed]
  67. Zang Y, Ran X, Yuan J, Wu H, Wang Y, Li H, et al. Genomic hallmarks and therapeutic targets of ribosome biogenesis in cancer. Brief Bioinform. 2024; 25: bbae023. [CrossRef] [Google scholar] [PubMed]
Journal Metrics
2024
CiteScore SJR SNIP
0.70.1470.167
Newsletter
Download PDF Supplementary File Download Full-Text XML Download Citation
0 0
Newsletter  

TOP