Pancreatic ductal adenocarcinoma (PDAC) is recognized as a highly aggressive fatal disease accounting for more than 90% of all pancreatic malignancies. Considering the limited effective treatment options and its low survival rate, studying PDAC's underlying mechanisms is of utmost importance. The present study focused on investigating PDAC expression data using WGCNA and Trader algorithms to shed light on the underlying mechanisms and identify the most reliable therapeutic candidates in PDAC. After analyzing a recently generated PDAC dataset (GSE132956), the obtained differentially expressed genes (DEGs) were subjected to different exploration steps. WGCNA was applied to cluster the co-expressed DEGs and found the disease's most correlated module and genes. The trader algorithm was utilized to analyze the constructed network of DEGs in STRING and identified the top 30 DEGs whose removal causes a maximum number of separate sub-networks. Hub genes were selected considering the lists of top identified nodes by the two algorithms. "Signaling by Rho GTPases," "Signaling by receptor tyrosine kinases," and "immune system" were top enriched gene ontology terms for the DEGs in the PDAC most correlated module and nine hub genes, including FYN, MAPK3, CDK2, SNRPG, GNAQ, PAK1, LPCAT4, MAP1LC3B, and FBN1 were identified by considering the top spotted DEGs by two algorithms. The findings provided evidence about the involvement of some pathways in the pathogenesis of PDAC and suggested several hub genes as therapeutic candidates via a comprehensive approach analyzing both the co-expression and PPI networks of DEGs in this cancer.
Research Article
Therapeutic Target Identification in Pancreatic Ductal Adenocarcinoma: A Comprehensive In-Silico Study employing WGCNA and Trader
https://doi.org/10.21203/rs.3.rs-2627227/v1
This work is licensed under a CC BY 4.0 License
Version 1
posted
You are reading this latest preprint version
Pancreatic ductal adenocarcinoma
therapeutic target
Network analysis
WGCNA
Trader algorithm
Pancreatic cancer (P.C.) is the 12th most common cancer and the 7th leading cause of cancer-related deaths, with a very low survival rate (5-year survival rate ~ 10%) (Luo, Tao et al. 2020, Mizrahi, Surana et al. 2020, Singhi and Wood 2021). Abnormal mutations, transcriptional reprogramming, and epigenetic and chromosomal alterations are known mechanisms in the pathogenesis of this malignancy (McGuigan, Kelly et al. 2018, Jagadeesan, Haran et al. 2021, Wood, Canto et al. 2022). Among the three types of pancreas tumors, pancreatic ductal adenocarcinoma (PDAC) is the most common one, accounting for more than 90% of all P.C. cases (Goral 2015, Lanfredini, Thapa et al. 2019, Jin and Bai 2020). Most PDACs arise from non-invasive precancerous lesions that progress to overt adenocarcinoma through a stepwise process (Hruban, Maitra et al. 2008, Mizrahi, Surana et al. 2020). Mutations in some genes (KRAS2, CDKN2A, TP53, and DPC4/SMAD4) and alterations in some pathways (RAS, cell-cycle control, DNA damage repair, and TGF-β3) are involved in this procedure (Feldmann, Beaty et al. 2007, Hidalgo 2010). Regarding PDAC treatment, removing the lesions via surgery is still the primary therapeutic approach with curative potential (McGuigan, Kelly et al. 2018, Zhao and Liu 2020). However, the risk of clinical morbidity and even death associated with pancreatic excision is significant (Canto, Kerdsirichairat et al. 2020). Hence, along with early diagnosis, an in-depth study of the inner regulatory mechanism of this malignancy is crucial to developing efficient treatments and enhancing the survival rate of patients.
Recently, systems biology has offered a holistic and interdisciplinary viewpoint to study biological systems in normal and disease states. As a well-known omics approach, expression profiling has been broadly used to recognize differential gene expression patterns in various conditions (Gholaminejad, Ghaeidamini et al. 2022). Once analyzed, such high-throughput data could provide valuable information about the disease’s molecular biology and be a great asset in identifying novel and efficient therapeutic targets (Gholaminejad, Gheisari et al. 2021, Roointan, Gheisari et al. 2021). Various bioinformatics tools and approaches have been developed to explore the data from expression profiles. Studying the co-expressed or regulatory networks of differentially expressed genes (DEGs) could be a valuable opportunity to identify the primary genes in the disease conditions.
The present experiment aimed to analyze and investigate a most recently generated PDAC transcriptomics dataset (GSE132956) using weighted gene co-expression networks analysis (WGCNA) and our previously developed algorithm called TRADER. In brief, after examination of the dataset using the linear model for microarray analysis (Limma) algorithm, the DEGs were subjected to different enrichment analyses, protein-protein interaction (PPI) and gene co-expression networks construction, and hub gene identification. Next, hub genes were explored throughout different steps, including expressional validation, single cell analysis, identifying potential drug candidates, and immune cell infiltration expressional correlation. The workflow of the present study is shown in Fig. 1.
Dataset analysis and PPI network construction
Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo/) was searched for the terms “pancreatic cancer” or “pancreatic adenocarcinoma.” The results were filtered by the selection of “human samples”, “expression by array,” and “published in the last year” items. Among the remained datasets, GSE132,956, including ten pancreatic ductal adenocarcinoma and five normal samples, was selected for the analysis. Principal component analysis (PCA) was utilized before the primary analysis to assess the dataset's quality and detect possible outliers. The network analyst online tool (http://www.networkanalyst.ca) was used for the quantile normalization procedure, removal of probes related to multiple genes, and performing the analysis using the Limma package. The identified DEGs (false discovery rate (FDR) cutoff < 0.049) were subjected to further investigation, including WGCNA and PPI network construction. All the identified DEGs were considered as seeds for the PPI network using the STRING database (confidence: 0.7) (Szklarczyk, Gable et al. 2019). The network was visualized and analyzed in Cytoscape software (version 3.9.1) (Shannon, Markiel et al. 2003), and the top DEGs with a high degree of centrality were identified.
Co-expression Network Construction And Analysis
The WGCNA algorithm in the R programming environment (version: 4.2.2) was utilized to construct the co-expressed networks and classify the DEGs into different co-expressed modules. At first, a matrix containing the intensity values of the identified DEGs was introduced to the algorithm. After the sample clustering step, the “pickSoftThreshold” feature was applied to discover the finest soft threshold power (from 1 to 20). The best soft threshold power (value: 12) was selected considering both the degree of independence and mean connectivity values. Afterward, several steps, including adjacency matrix construction, topological overlap matrix (TOM) estimation, construction of modules, and dynamic branch cutting (merging threshold of 0.15), were carried out consecutively to have the co-expressed modules. Furthermore, a TOM dissimilarity-based heatmap was constructed to illustrate the gene adjacencies and verify the module's division accuracy in the identified modules. After defining the group of samples as either control or cancer, the “labeledHeatmap” function was applied to identify the most cancer-correlated module/s.
Trader optimization algorithm for analyzing the PPI network
The Trader algorithm was utilized to identify driver nodes in the constructed protein-protein interaction network associated with pancreatic cancer. This algorithm randomly creates several potential answers and divides them into groups by chance. Every candidate solution (C.S.) shows some proteins as the pathogenesis of pancreatic cancer. However, the selected proteins are not presumably effective in inducing cancer. Therefore, the algorithm tries to change the C.S.s and generate better potential answers based on three operators. The first operator, called distributing and formulated in Eq. 1, chooses several variables of the best of a group and passes them to the other C.S.s of the same group.
R, i, and m represent a randomly selected subset of the variables, the ith C.S., and the mth C.S. (the best of the group) of a group, respectively.
The second operator, named retailing and coded in Eq. 2, aims to induce small changes in a C.S.
R, K, and rand show a randomly selected subset of the variables, a random floating-point value between − 1 and 1, and a function generating a random floating-point value between the determined range, respectively. The third operator, called Importing-exporting/I.E. and formulated in Eq. 3, transfers values of the best potential solutions of the groups to the variables of a C.S. of interest.
Like Eq. 1 and Eq. 2, R indicates a randomly selected subset of the variables. Also, \({CS}_{j}^{{G}_{I}}\) and \({CS}_{j}^{{G}_{E}}\) show the jth variables of the importer and exporter C.S.s (the best of their related groups), respectively. After applying the operators mentioned above to the potential answers, the fitness/score function is called. In identifying candidate driver nodes of the pancreatic cancer-related PPI network, the score function removes the selected proteins/nodes and their edges/interactions from the networks and counts the total number of disjoint networks using the depth first search (DFS) algorithm. The higher generated subnetworks mean the higher worthiness of the selected proteins. Of note, in the Trader algorithm, the received changes will be ignored if they have no improvement in the C.S. score. The proposed algorithm was
implemented in MATLAB programming language (MATLAB R2022a).
Selection Of Candidate Hub Genes
The candidate hub genes were selected by analyzing the constructed PPI network and co-expression networks. At first, considering module membership (kME) and gene significance (G.S.) values (kME: >0.8, G.S.: >0.6), a gene list was extracted from the most disease correlated module. In parallel, the edge list of the constructed PPI network was introduced to Trader to identify the top 30 DEGs in the network. As mentioned above, in a PPI network, Trader can recognize nodes whose removal causes a maximum number of separate sub-networks (Masoudi-Sobhanzadeh, Gholaminejad et al. 2022). Finally, the intersection of genes in the obtained lists by two algorithms (Trader and WGCNA) were identified as hub genes in the disease condition.
Functional Enrichment Analysis
To understand the underlying molecular mechanisms and pathways in the pancreatic cancer condition, gene ontology (G.O.) and Reactome pathway enrichment analyses were performed on the genes of the module of interest. The same procedure was carried out for all the DEGs in other co-expressed modules. The CluGO module (version 2.5.9) (Bindea, Mlecnik et al. 2009) in Cytoscape software was utilized to perform the enrichment analysis (significant enrichment threshold: P < 0.05). The parent terms were identified and presented as the main terms.
Hub-gene Studies
In this step, the identified hub genes were subjected to various in silico experiments to understand their functions and roles in pancreatic cancer. At first, Gene Set Cancer Analysis (GSCA) platform was utilized to validate the expression of the hub genes in pancreatic vs. normal samples. The expressional validation analysis was performed on 179 − 171 paired tumor and normal samples. Then, TIMER, a known platform for analyzing immune cell infiltration in different cancer types, was utilized to investigate whether the expression levels of the hub genes are associated with immune cell infiltration in pancreatic cancer (Li, Fu et al. 2020). Next, to find the potential drug candidates for targeting the identified hub genes in pancreatic cancer, DGIdb (Drug–Gene Interaction Database) was utilized (Freshour, Kiwala et al. 2021).
DEGs identification, network construction, and pathway analysis: 4131 significant DEGs were identified
Before the dataset analysis, some preprocessing steps, including PCA, and normalizing methods, were carried out to guarantee the accuracy of the core analysis. At first, the quality of the dataset was assured by the PCA. According to the plot, there was no outlier among the samples, and both groups (controls and cases) were separated based on PC1 (Fig. 2A). Moreover, quantile normalization was done to ensure that each sample's expression distributions were the same over the whole dataset. Next, the core analysis identified 4131 significant DEGs, including 2128 down-regulated and 2003 up-regulated genes. Figure 2B shows a volcano plot representing the top identified DEGs based on the log2FC. The analysis result, including all the names and values of the identified DEGs, are provided in a supplementary file (Supplementary file 1 – Dataset analysis result). A PPI network with 2166 nodes and 11485 edges was constructed and analyzed in the next step. Figure 2C shows the top 30 DEGs in the network based on the degree of centrality. To understand the main biological pathways in which DEGs are involved, Reactome pathway enrichment analysis was performed, and the main terms are shown in Fig. 2D.
Co-expression analysis: Co-expressed DEGs were clustered into 7 modules; the blue module was the most correlated disease.
The results of co-expression analysis using the WGCNA algorithm, including the scale independence and mean connectivity plots, cluster dendrogram, eigengene dendrogram and adjacency heatmap, network heatmap plot, module-trait relationship heatmap, and module membership vs. gene significance plot for the blue module, all are shown in a multi-panel figure (Fig. 3A-F). Based on scale independence and mean connectivity results, a soft threshold of 12 was designated to get an approximate scale-free topology (Fig. 3A). Additionally, after performing the functions related to the hierarchical clustering and module merging, DEGs were grouped into seven distinct co-expression modules (Royal blue, blue, cyan, black, brown, light green, and turquoise) (Fig. 3B). The resulting clustering dendrogram and eigengene adjacency heatmap displayed the division of the established modules into two clusters (Fig. 3C). The network heatmap plot revealed the topological overlap adjacency among genes across modules (Fig. 3D). According to the results of the module-trait relationship plot, co-expressed genes in the blue module had the highest correlation to the disease state and were selected for further analysis (Fig. 3E). Considering the kME and G.S. values of the co-expressed DEGs in the blue module, 580 DEGs were identified as the top DEGs in the module (Fig. 3F). The names and values related to the clustered DEGs in the co-expressed modules are provided in supplementary file 2 (Supplementary file 2 - WGCNA and Trader results).
G.O. enrichment and pathway analysis: blue module genes were mainly enriched in Signaling by Rho GTPases, Signaling by receptor tyrosine kinases, and immune system
Gene ontology enrichment analysis was performed for the co-expressed genes in all the individual modules. The top biological process terms were presented for each module (Fig. 4A). As the disease's most correlated module, the genes in the blue module were subjected to more enrichment analyses. According to the Reactome pathway analysis results, the blue module genes were mainly enriched in “Signaling by Rho GTPases”, “Signaling by receptor tyrosine kinases”, “EPH-Ephrin signaling”, “extracellular matrix organization”, “hemostasis” and “immune system” (Fig. 4B). The results of the biological process terms were in line with the Reactome analysis results, and the main enriched terms were “Rho protein signal transduction”, “small GTPase mediated signal transduction”, “actin cytoskeleton organization”, and several immune system-related terms. Table 1 lists the enriched G.O. terms for the blue module genes.
| G.O. | Term | G.O. term | Nr. Genes | % Associated Genes | Term PValue Corrected with Bonferroni step down |
|---|---|---|---|---|---|
| B.P. | negative regulation of ERBB signaling pathway | GO:1901185 | 10 | 31.25 | 0.022547045 |
| antigen processing and presentation of peptide antigen | GO:0048002 | 16 | 23.188406 | 0.005346598 | |
| regulation of Rho protein signal transduction | GO:0035023 | 19 | 18.811882 | 0.017618911 | |
| response to interferon-gamma | GO:0034341 | 26 | 17.567568 | 0.001804067 | |
| regulation of cell-substrate adhesion | GO:0010810 | 39 | 16.810345 | 1.45064E-05 | |
| Rho protein signal transduction | GO:0007266 | 26 | 16.666666 | 0.004977302 | |
| actin cytoskeleton organization | GO:0030036 | 97 | 12.548512 | 1.96237E-08 | |
| response to wounding | GO:0009611 | 73 | 11.793215 | 0.000138442 | |
| small GTPase mediated signal transduction | GO:0007264 | 65 | 11.711712 | 0.000901925 | |
| blood vessel development | GO:0001568 | 86 | 11.466666 | 2.62259E-05 | |
| M.F. | cell adhesion molecule binding | GO:0050839 | 78 | 13.04348 | 5.80E-08 |
| cytoskeletal protein binding | GO:0008092 | 108 | 10.24668 | 2.39E-05 | |
| enzyme binding | GO:0019899 | 204 | 9.255898 | 1.15E-07 | |
| identical protein binding | GO:0042802 | 196 | 9.032258 | 2.38E-06 | |
| metalloendopeptidase activity involved in amyloid precursor protein catabolic process | GO:1902945 | 4 | 80 | 0.024879 | |
| negative regulation of catalytic activity | GO:0043086 | 93 | 11.07143 | 5.49E-06 | |
| protein domain-specific binding | GO:0019904 | 74 | 10.10929 | 0.006617 | |
| regulation of hydrolase activity | GO:0051336 | 113 | 10.38603 | 5.54E-06 | |
| CC | Arp2/3 protein complex | GO:0005885 | 7 | 70 | 8.23E-05 |
| MHC protein complex | GO:0042611 | 8 | 28.571428 | 0.046645951 | |
| immunological synapse | GO:0001772 | 13 | 27.659575 | 6.80E-04 | |
| phagocytic vesicle | GO:0045335 | 25 | 16.891891 | 5.98E-04 | |
| cell-substrate junction | GO:0030055 | 78 | 16.595745 | 2.87E-14 | |
| lamellipodium | GO:0030027 | 33 | 14.537445 | 7.72E-04 | |
| ruffle | GO:0001726 | 27 | 13.432836 | 0.018817795 | |
| membrane raft | GO:0045121 | 47 | 12.912087 | 1.46E-04 | |
| collagen-containing extracellular matrix | GO:0062023 | 56 | 12.043011 | 1.33E-04 | |
| actin cytoskeleton | GO:0015629 | 62 | 11.63227 | 1.09E-04 |
Hub gene identification: 9 DEGs were spotted by both the Trader and WGCNA algorithms
In this stage, we applied two algorithms with different strategies to reach several candidate hub genes in the pancreatic adenocarcinoma condition. The first strategy was based on the PPI network, where the Trader algorithm identified the top 30 DEGs. The second was based on co-expression network analysis, where WGCNA presented the top co-expressed DEGs in cancer's most correlated module. The intersection between the two lists was nine DEGs, including FYN, MAPK3, CDK2, SNRPG, GNAQ, MAP1LC3B, LPCAT4, PAK1, and FBN1 (Table 2) (Fig. 5A). The top DEGs by both strategies are provided in supplementary file 2 (Supplementary file 2 - WGCNA and Trader results).
| Gene symbol | Gene description | Role in cancer pathogenesis | Proposed (Approved) Drug/s | Top 10% in the PPI network (degree centrality) |
|---|---|---|---|---|
| FYN | Tyrosine-protein kinase | Reduction of FYN blocked cell proliferation, migration, and invasion in the MDA-MB-231 cells (Xie, Yu et al. 2016). Inhibition of FYN in mouse glioma models delays tumor initiation and progression (Comba, Dunn et al. 2020) | DASATINIB PAZOPANIB | YES |
| MAPK3 | MAP kinase-activated protein kinase 3 | Targeting MAPK3 by miR-143 inhibited the proliferation of the breast cancer cells (Du, Zhang et al. 2020) | ETOPOSIDE CYCLOPHOSPHAMIDE SORAFENIB | YES |
| CDK2 | Cyclin-dependent kinase 2 | CDK2 and CDK4/6 inhibition neutralizes palbociclib resistance in breast cancer (Pandey, Park et al. 2020) | PACLITAXEL CARBOPLATIN ERIBULIN DEXAMETHASONE | YES |
| SNRPG | Small nuclear ribonucleoprotein polypeptide G | SNRPG downregulation results in cell cycle arrest and sensitizing human glioblastoma cells to temozolomide (Lan, Lou et al. 2020) | - | YES |
| GNAQ | Guanine nucleotide-binding protein G(q) subunit alpha | GNAQ knockdown results in a blockage in the cell cycle and disturb the proliferation of gastric cancer cells (Lai, Zeng et al. 2019) | EVEROLIMUS | YES |
| PAK1 | Serine/threonine-protein kinase | PAK1 inhibition suppresses pancreatic cancer through the down-regulation of PD-L1 (Wang, Zhan et al. 2020) | - | YES |
| LPCAT4 | Lysophospholipid acyltransferase | LPCAT4 is linked to intestinal stem cell proliferation and tumorigenesis (Butler, Perone et al. 2020) | - | No |
| MAP1LC3B | Microtubule-associated protein 1 light chain 3B | MAP1LC3B up-regulation is linked to breast cancer and melanoma progression (Lazova, Camp et al. 2012, Zhao, Yang et al. 2013) | - | No |
| FBN1 | Fibrillin-1 | FBN1 knock-down inhibited xenograft (papillary thyroid) tumor formation in vivo (Ma, Wei et al. 2016) FBN1 knockdown inhibits cell migration and invasion in ovarian cancer (Wang, Liu et al. 2015) | - | No |
Hub Gene Examination: Expressional Validation, Single Cell Analysis, And Immune Cell Infiltration Study
The results of expressional analysis of the identified hub genes in the GSCA database confirmed the up-regulation of all 9 hub genes in the pancreatic samples (Fig. 5B). Such findings validated the expressional analysis process in the present experiment. Using the TIMER database, the correlation of each hub gene with the infiltration of different immune cells was assessed. According to the TIMER data, there is a positive correlation between the expression of several hub genes (MAP1LC3B, GANQ, FYN, and FBN1) and infiltration of B-cells, CD8 + T cells, CD4 + T cells, macrophages, Neutrophils, and Dendritic cells (Fig. 6). Moreover, the currently approved drugs targeting the hub genes were identified using the DGIdb database. Among the 9 hub genes, FYN, MAPK3, CDK2, and GNAQ were found to be targets of several drugs like Dasatinib, Pazopanib, Etoposide, Cyclophosphamide, Sorafenib, Paclitaxel, Carboplatin, Eribulin, Dexamethasone, and Everolimus (Table 2). Moreover, the results of our literature review revealed the involvement of almost all the identified hub genes in the progression of different cancers (Table 2).
Targeted therapy has been the focus of recent investigations, leading to substantial progress in identifying novel therapeutic targets in cancers (Kroes, Jastrow et al. 2000, Lee, Tan et al. 2018, Li, Ren et al. 2021). Apart from the known classical biology approaches, the systems biology view has recently offered more holistic and cost-effective strategies for exploring cancer-related biological systems, understanding the pathogenicity, and identifying novel therapeutic targets (Yan, Risacher et al. 2018, Veenstra 2021).
Generally, biological systems are represented as networks, which are groups of binary interactions among different entities (Green, Şerban et al. 2018, Liu, Ma et al. 2020). The nodes of such networks could be the differentially expressed biological entities (e.g., genes, proteins, metabolites, or non-coding RNAs) in a specific condition like cancer. In this case, constructing, analyzing, and translating such networks could enhance our understanding of cancer biology and lead us to the key players in its pathogenicity (Buphamalai, Kokotovic et al. 2021, Prasad, AlOmar et al. 2021, Tomkins and Manzoni 2021). Considering this topic, so far, different algorithms have been developed and applied for analyzing biological networks (Das, Alphonse et al. 2021, Hu, Zhang et al. 2021, Dilmaghani, Brust et al. 2022). WGCNA, for instance, is one of the well-known algorithms for co-expressional network analysis of high-throughput expressional data (Langfelder and Horvath 2008). The algorithm could cluster the co-expressed genes into distinct modules, identify the most disease-correlated module and finally recognize the key genes in that module (Giulietti, Righetti et al. 2018). Up to now, by applying this algorithm, several potential therapeutic targets have been introduced in various disease conditions (Giulietti, Righetti et al. 2018, Ding, Li et al. 2019, Gholaminejad, Fathalipour et al. 2021, Gholaminejad, Roointan et al. 2021, Yin, Wang et al. 2021, Huang, Tong et al. 2022).
Topological analysis of PPI networks is another common strategy to identify the key nodes (as potential therapeutic targets) in disease networks (Theodosiou, Efstathiou et al. 2017, Nowakowska and Kotulska 2022). According to this approach, centrality features of nodes like degree, betweenness, closeness, eigenvector, etc., are usually considered to identify the key nodes in the PPI networks (Sharma, Bhattacharyya et al. 2016, Sanchez and Mackenzie 2020). In this regard, the topmost centrality-based nodes are considered possible therapeutic targets, as they have more interactions with other nodes in the network. However, as a major limitation of this strategy, some essential nodes with low centrality values might be ignored (Ren, Wang et al. 2015, Li, Zeng et al. 2022). Accordingly, more accurate and reliable strategies are needed in analyzing biological networks to identify key elements. Recently, our team has developed and tested a novel algorithm called Trader to identify key nodes in the PPI networks (Masoudi-Sobhanzadeh, Gholaminejad et al. 2022). The Trader’s mechanism of action is based on detecting a minimal set of nodes whose elimination creates a maximum number of sub-networks (Masoudi-Sobhanzadeh, Gholaminejad et al. 2022). Accordingly, the identified nodes could be considered more reliable therapeutic target candidates
In the present integrative systems biology experiment, we applied the WGCNA and Trader algorithms to analyze a PDAC expression dataset and identify the potential therapeutic targets for treating this cancer. Moreover, the PDAC underlying molecular pathways and biological processes have been explored.
The enrichment analysis was conducted by evaluating the clustered DEGs in the PDAC most correlated module (blue module). The results revealed a significant involvement of Rho GTPases signaling networks, signaling by receptor tyrosine kinases, and immune system pathways in PDAC tumors. Previously, the involvement of immune-related pathways has been shown in the pathogenesis of PDAC (Elebo, Fru et al. 2020). In addition, different immune cells, including CD8+, CD4 + T cells, dendritic cells, and natural killer cells, are shown to be active in the microenvironment of PDAC, inhibiting tumor growth and progression (Tjomsland, Sandström et al. 2010). Notably, pancreatic adenocarcinoma features a highly immunosuppressive microenvironment applying different cells and mechanisms to circumvent the immune responses (Haqq, Howells et al. 2014, Karamitopoulou 2019). Based on previous investigations, a dense desmoplastic stroma, cancer-associated fibroblasts (CAFs), programmed death-ligand 1 (PD-L1), and JAK/STAT signaling pathway are known to play a role in the host immune response evasion (Spranger and Gajewski 2018). It seems that targeting such molecules/pathways could be a possible approach to sensitize pancreatic cancer to immunotherapies (Deng, Xia et al. 2021).
Signaling by Rho GTPases was another enriched term considering the clustered DEGs in the blue module. Such a result might not be irrelevant to the therapy resistance nature of the PDAC since the Rho GTPase signaling has been shown to enhance a plethora of oncogenic microRNAs and pro-survival molecules in various cancer conditions. In normal conditions, Rho GTPases are key players in the regulation of cytoskeleton dynamics, cell cycle, cell survival, cell adhesion, and cell migration (Zubor, Dankova et al. 2020, Rodenburg and van Buul 2021). In cancer conditions, Rho GTPases have been shown to play an essential role in the invasive behavior of cancer cells, like metastasis formation and extravasation (Rodenburg and van Buul 2021). Targeting the members of Rho GTPase signaling networks in cancers might be an actual strategy to constrain extravasation and metastasis. In PDAC tumors, Rho GTPases are the central downstream regulators of K-Ras mutant forms, found in more than 90% of cases (Ryan, Hong et al. 2014). Some investigations have previously shown the connection between the Rho GTPase signaling networks and the invasiveness of PDAC tumors (Taniuchi, Nakagawa et al. 2005, Kimmelman, Hezel et al. 2008, Melzer, Hass et al. 2017, Rane and Minden 2019). Rio Kinase 3 (RIOK3) and p21 Activated Kinase 4 (PAK4), for instance, are two examples of Rho-binding proteins promoting pancreas ductal cell motility and invasion through binding to Rho GTPases (Kimmelman, Hezel et al. 2008). The result of a recent experiment on pancreatic cancer metastasis also revealed the importance of Rho GTPases in the metastatic behavior of cancer cells. Silencing of cdc42 and Rac1 as two members of this family was shown to reduce the enhanced tumor cell motility under applied mechanical stress, resembling the active state of the tumor (Kalli, Li et al. 2022). Despite this data, only a small number of drugs blocking Rho GTPase signaling are currently being tested in clinical settings to treat cancer (Crosas-Molist, Samain et al. 2022). Consistent with the available information, our findings could also be a witness to the therapeutic potential of Rho GTPases in PDAC tumors.
In the next step of this work, 9 hub genes were identified by applying an integrative approach considering the lists of top DEGs determined by both WGCNA and Trader. As a result, FYN, MAPK3, CDK2, SNRPG, GNAQ, PAK1, LPCAT4, MAP1LC3B, and FBN1 were the intersection of the identified list of DEGs by both algorithms. The identified hubs could be the focus of more investigations considering their potential as therapeutic targets in PDAC tumors. Of note, PAK1, as a serine/threonine-protein kinase, and MAPK3, as a member of the mitogen-activated protein kinase (MAP kinase) family, are two well-known effectors of Rho GTPase signaling networks. The PAKs are the primary GTPase-regulated kinases to be recognized, and the PAK1 specifically was shown to interact with different Rho GTPases like rac1, rac2, rac3, and cdc45 (Manser, Leung et al. 1994, Mira, Benard et al. 2000, Chetty, Ha et al. 2022). Besides its association with the Rho GTPases and its crucial regulatory role in cytoskeletal dynamics, PAK1 is involved in various oncogenic pathways (Yao, Li et al. 2020, Guo, Liu et al. 2022). MAPK3 (ERK1), another kinase among the identified hub genes, was shown to have an essential role in regulating cell proliferation, differentiation, and survival. Targeting this kinase in different cancers, including breast cancer (Du, Zhang et al. 2020), prostate cancer (Nickols, Nazarian et al. 2019), multiple myeloma (Adamia, Bhatt et al. 2022) and glioma (Ku, Edes et al. 2016) showed promising outcomes.
Apart from FYN, MAPK3, CDK2, and PAK1, which are functioning as kinases, other identified hub genes were FBN1, an essential protein in extracellular matrix regulation, MAP1LC3B, involving in autophagic vacuole formation, LPCAT4, as an acyltransferase, SNRPG, as a core component of spliceosome, and GNAQ, as a transmembrane signaling transducer. More details about the identified hub genes and their potential as therapeutic targets are provided in Table 2.
Further analysis of the hub genes revealed a positive correlation of several hub genes with the infiltration of the immune cells. Infiltrating immune cells are widely established to be part of the tumor microenvironment and crucial for tumor development, invasion, and metastasis (Huang, Chen et al. 2020, Niu, Yan et al. 2020). According to reports, tumor-infiltrating immune cells could enhance and depress antitumor immunity in the tumor microenvironment (Huang, Chen et al. 2020). Thus, it is critical to identify the status of different immune cell infiltration in the tumor microenvironment. The obtained results in this step provided evidence about the involvement of MAP1LC3B, GANQ, FYN, and FBN1 in the infiltration of immune cells, including B-cells, CD8 + T cells, CD4 + T cells, macrophages, Neutrophils, Dendritic cells. Such findings suggested the potential involvement of the mentioned genes in establishing the PDAC immune microenvironment.
In conclusion, by applying two algorithms investigating both the co-expressional and PPI networks of the PDAC DEGs, the obtained results suggested the therapeutic potential of the Rho GTPase signaling members and several hub genes in PDAC tumors. The proposed pathways/targets could be the focus of in vitro and in vivo investigations aiming to treat pancreatic cancer.
A critical requirement for clinical pharmaceutics is identifying novel and effective treatment targets, particularly for complex disorders like cancers. Considering the limited effective treatment options for pancreatic ductal adenocarcinoma, the study of its underlying mechanisms aiming for the discovery of novel therapeutic targets is of utmost importance. So far, various bioinformatic tools and approaches have been developed to examine cancer-related high-throughput data. In this case, exploring either the co-expressed or regulatory networks of differentially expressed genes (DEGs) via novel strategies could be helpful for the identification of the key genes in cancer conditions. Notably, the discovery of therapeutic targets in such networks could have a significant impact on the attrition rates in the drug development pipeline. The present study applied a novel integrative strategy using a previously developed algorithm (WGCNA) and a newly developed algorithm called Trader to analyze and explore the network of DEGs in PDAC malignancy. The two main aims of the present study are understanding the underlying molecular mechanisms and identifying the most reliable potential targets in PDAC. After preprocessing and analyzing a recently generated PDAC transcriptomics dataset, the PPI network of the identified DEGs was introduced to Trader to identify the most subset of genes whose removal generates the maximum number of sub-networks in the whole PPI network.
On the other hand, the intensity values of the DEGs were entered into the WGCNA algorithm to identify the disease most correlated module and the top DEGs in that module. The DEGs spotted by the two algorithms, including FYN, MAPK3, CDK2, SNRPG, GNAQ, PAK1, LPCAT4, MAP1LC3B, and FBN1, were identified as the most reliable therapeutic candidates in PDAC. The results of G.O. enrichment revealed the involvement of Rho GTPase signaling networks, signaling by receptor tyrosine kinases, and immune system-related pathways in the pathogenesis of PDAC. The identified driver nodes and the members of the enriched pathways could be the focus of more investigations aiming to hinder the progression of PDAC. Moreover, the applied comprehensive approach utilizing the co-expression and PPI network of DEGs using WGCNA and Trader could be inspirational for both the bioinformaticians and biologists in this area of research.
Acknowledgments
We thank members of the Regenerative Medicine Research Center, especially Mr. Mohammadjavad Naghdibadi, for their help in some parts of this work.
Statement of ethics:
An ethics statement was not required for this study type, and no human or animal subjects or materials were used.
Conflict of interest statement
The authors declare that they have no conflict of interest.
Funding sources
The authors received no funds for this work.
Authors' contributions
Amir Roointan and Parvin Yavari participated in the study design, analysis of the dataset using the WGCNA algorithm, interpretation of the results, and manuscript drafting. Yosef Masoudi-Sobhanzadeh participated in the study design, interpreting the result and drafting the manuscript. The TRADER algorithm also was applied by Yosef Masoudi-Sobhanzadeh. All authors reviewed the final version of the manuscript.
Data availability statement
The analyzed dataset by the present study is available in the GEO repository, [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE132956]
- Adamia, S., S. Bhatt, K. Wen, Z. Chyra, G. G. Fell, Y.-T. Tai, M. S. Pioso, I. Abiatari, A. Letai and D. M. Dorfman (2022). "Combination therapy targeting Erk1/2 and CDK4/6i in relapsed refractory multiple myeloma." Leukemia 36(4): 1088-1101.
- Bindea, G., B. Mlecnik, H. Hackl, P. Charoentong, M. Tosolini, A. Kirilovsky, W.-H. Fridman, F. Pagès, Z. Trajanoski and J. Galon (2009). "ClueGO: a Cytoscape plug-in to decipher functionally grouped gene ontology and pathway annotation networks." Bioinformatics 25(8): 1091-1093.
- Buphamalai, P., T. Kokotovic, V. Nagy and J. Menche (2021). "Network analysis reveals rare disease signatures across multiple levels of biological organization." Nature communications 12(1): 1-15.
- Butler, L. M., Y. Perone, J. Dehairs, L. E. Lupien, V. de Laat, A. Talebi, M. Loda, W. B. Kinlaw and J. V. Swinnen (2020). "Lipids and cancer: Emerging roles in pathogenesis, diagnosis and therapeutic intervention." Advanced drug delivery reviews 159: 245-293.
- Canto, M. I., T. Kerdsirichairat, C. J. Yeo, R. H. Hruban, E. J. Shin, J. A. Almario, A. Blackford, M. Ford, A. P. Klein and A. A. Javed (2020). "Surgical outcomes after pancreatic resection of screening-detected lesions in individuals at high risk for developing pancreatic cancer." Journal of Gastrointestinal Surgery 24(5): 1101-1110.
- Chetty, A. K., B. H. Ha and T. J. Boggon (2022). "Rho family GTPase signaling through type II p21-activated kinases." Cellular and Molecular Life Sciences 79(12): 1-16.
- Comba, A., P. J. Dunn, A. E. Argento, P. Kadiyala, M. Ventosa, P. Patel, D. B. Zamler, F. J. Núñez, L. Zhao and M. G. Castro (2020). "Fyn tyrosine kinase, a downstream target of receptor tyrosine kinases, modulates antiglioma immune responses." Neuro-oncology 22(6): 806-818.
- Crosas-Molist, E., R. Samain, L. Kohlhammer, J. L. Orgaz, S. L. George, O. Maiques, J. Barcelo and V. Sanz-Moreno (2022). "Rho GTPase signaling in cancer progression and dissemination." Physiological Reviews 102(1): 455-510.
- Das, M., P. Alphonse and S. Kamalanathan (2021). Markov Clustering Algorithms and Their Application in Analysis of PPI Network of Malaria Genes. 2021 11th IEEE International Conference on Intelligent Data Acquisition and Advanced Computing Systems: Technology and Applications (IDAACS), IEEE.
- Deng, Y., X. Xia, Y. Zhao, Z. Zhao, C. Martinez, W. Yin, J. Yao, Q. Hang, W. Wu, J. Zhang, Y. Yu, W. Xia, F. Yao, D. Zhao, Y. Sun, H. Ying, M. C. Hung and L. Ma (2021). "Glucocorticoid receptor regulates PD-L1 and MHC-I in pancreatic cancer cells to promote immune evasion and immunotherapy resistance." Nat Commun 12(1): 7041.
- Dilmaghani, S., M. R. Brust, C. H. Ribeiro, E. Kieffer, G. Danoy and P. Bouvry (2022). "From communities to protein complexes: A local community detection algorithm on PPI networks." Plos one 17(1): e0260484.
- Ding, M., F. Li, B. Wang, G. Chi and H. Liu (2019). "A comprehensive analysis of WGCNA and serum metabolomics manifests the lung cancer‐associated disordered glucose metabolism." Journal of Cellular Biochemistry 120(6): 10855-10863.
- Du, Y., J. Zhang, Y. Meng, M. Huang, W. Yan and Z. Wu (2020). "MicroRNA-143 targets MAPK3 to regulate the proliferation and bone metastasis of human breast cancer cells." Amb Express 10(1): 1-8.
- Elebo, N., P. Fru, J. Omoshoro‑Jones, G. Patrick Candy and E. E. Nweke (2020). "Role of different immune cells and metabolic pathways in modulating the immune response in pancreatic cancer." Molecular Medicine Reports 22(6): 4981-4991.
- Feldmann, G., R. Beaty, R. H. Hruban and A. Maitra (2007). "Molecular genetics of pancreatic intraepithelial neoplasia." Journal of Hepato‐Biliary‐Pancreatic Surgery 14(3): 224-232.
- Freshour, S. L., S. Kiwala, K. C. Cotto, A. C. Coffman, J. F. McMichael, J. J. Song, M. Griffith, O. L. Griffith and A. H. Wagner (2021). "Integration of the Drug–Gene Interaction Database (DGIdb 4.0) with open crowdsource efforts." Nucleic acids research 49(D1): D1144-D1151.
- Gholaminejad, A., M. Fathalipour and A. Roointan (2021). "Comprehensive analysis of diabetic nephropathy expression profile based on weighted gene co-expression network analysis algorithm." BMC nephrology 22(1): 1-13.
- Gholaminejad, A., M. Ghaeidamini, J. Simal-Gandara and A. Roointan (2022). "An Integrative in silico Study to Discover Key Drivers in Pathogenicity of Focal and Segmental Glomerulosclerosis." Kidney and Blood Pressure Research: 1-13.
- Gholaminejad, A., Y. Gheisari, S. Jalali and A. Roointan (2021). "Comprehensive analysis of IgA nephropathy expression profiles: identification of potential biomarkers and therapeutic agents." BMC nephrology 22(1): 1-10.
- Gholaminejad, A., A. Roointan and Y. Gheisari (2021). "Transmembrane signaling molecules play a key role in the pathogenesis of IgA nephropathy: a weighted gene co-expression network analysis study." BMC immunology 22(1): 1-17.
- Giulietti, M., A. Righetti, G. Principato and F. Piva (2018). "LncRNA co-expression network analysis reveals novel biomarkers for pancreatic cancer." Carcinogenesis 39(8): 1016-1025.
- Goral, V. (2015). "Pancreatic cancer: pathogenesis and diagnosis." Asian Pacific Journal of Cancer Prevention 16(14): 5619-5624.
- Green, S., M. Şerban, R. Scholl, N. Jones, I. Brigandt and W. Bechtel (2018). "Network analyses in systems biology: new strategies for dealing with biological complexity." Synthese 195(4): 1751-1777.
- Guo, P., Y. Liu, J. Feng, S. Tang, F. Wei and J. Feng (2022). "p21-activated kinase 1 (PAK1) as a therapeutic target for cardiotoxicity." Archives of Toxicology: 1-20.
- Haqq, J., L. M. Howells, G. Garcea, M. S. Metcalfe, W. P. Steward and A. R. Dennison (2014). "Pancreatic stellate cells and pancreas cancer: current perspectives and future strategies." European journal of cancer 50(15): 2570-2582.
- Hidalgo, M. (2010). "Pancreatic cancer." New England Journal of Medicine 362(17): 1605-1617.
- Hruban, R. H., A. Maitra and M. Goggins (2008). "Update on pancreatic intraepithelial neoplasia." International journal of clinical and experimental pathology 1(4): 306.
- Hu, L., J. Zhang, X. Pan, X. Luo and H. Yuan (2021). "An effective link-based clustering algorithm for detecting overlapping protein complexes in protein-protein interaction networks." IEEE Transactions on Network Science and Engineering 8(4): 3275-3289.
- Huang, C., Q. Tong, W. Zhang, X. Chen and Z. Pan (2022). "WGCNA Reveal Potential Diagnosis Biomarkers and Therapeutic Targets for COVID-19 Infection in Patients with Sepsis."
- Huang, L., H. Chen, Y. Xu, J. Chen, Z. Liu and Q. Xu (2020). "Correlation of tumor‐infiltrating immune cells of melanoma with overall survival by immunogenomic analysis." Cancer medicine 9(22): 8444-8456.
- Jagadeesan, B., P. H. Haran, D. Praveen, P. R. Chowdary and M. V. Aanandhi (2021). "A comprehensive review on pancreatic cancer." Res J Pharm Technol 14: 552-554.
- Jin, C. and L. Bai (2020). "Pancreatic cancer–current situation and challenges." Gastroenterol Hepatol Lett 2(1): 1-3.
- Kalli, M., R. Li, G. B. Mills, T. Stylianopoulos and I. K. Zervantonakis (2022). "Mechanical Stress Signaling in Pancreatic Cancer Cells Triggers p38 MAPK-and JNK-Dependent Cytoskeleton Remodeling and Promotes Cell Migration via Rac1/cdc42/Myosin IIMechanical Stress–Induced Adaptation in Pancreatic Cancer Cells." Molecular Cancer Research: OF1-OF13.
- Karamitopoulou, E. (2019). "Tumour microenvironment of pancreatic cancer: immune landscape is dictated by molecular and histopathological features." British journal of cancer 121(1): 5-14.
- Kimmelman, A. C., A. F. Hezel, A. J. Aguirre, H. Zheng, J.-h. Paik, H. Ying, G. C. Chu, J. X. Zhang, E. Sahin and G. Yeo (2008). "Genomic alterations link Rho family of GTPases to the highly invasive phenotype of pancreas cancer." Proceedings of the National Academy of Sciences 105(49): 19372-19377.
- Kroes, R. A., A. Jastrow, M. G. McLone, H. Yamamoto, P. Colley, D. S. Kersey, V. W. Yong, E. Mkrdichian, L. Cerullo and J. Leestma (2000). "The identification of novel therapeutic targets for the treatment of malignant brain tumors." Cancer letters 156(2): 191-198.
- Ku, M.-C., I. Edes, I. Bendix, A. Pohlmann, H. Waiczies, T. Prozorovski, M. Günther, C. Martin, S. A. Wolf and H. Kettenmann (2016). "ERK1 as a Therapeutic Target for Dendritic Cell Vaccination against High-Grade GliomasTargeting ERK1 in DC Vaccines for Glioma." Molecular Cancer Therapeutics 15(8): 1975-1987.
- Lai, J., X. Zeng, X. Yu and J. Ye (2019). "GNAQ AFFECTS THE OCCURRENCE AND DEVELOPMENT OF GASTRIC CANCER THROUGH THE P53/P21 AND MEK/ERK PATHWAYS." ACTA MEDICA MEDITERRANEA 35(4): 2139-2143.
- Lan, Y., J. Lou, J. Hu, Z. Yu, W. Lyu and B. Zhang (2020). "Downregulation of SNRPG induces cell cycle arrest and sensitizes human glioblastoma cells to temozolomide by targeting Myc through a p53-dependent signaling pathway." Cancer biology & medicine 17(1): 112.
- Lanfredini, S., A. Thapa and E. O'Neill (2019). "RAS in pancreatic cancer." Biochemical Society Transactions 47(4): 961-972.
- Langfelder, P. and S. Horvath (2008). "WGCNA: an R package for weighted correlation network analysis." BMC bioinformatics 9(1): 1-13.
- Lazova, R., R. L. Camp, V. Klump, S. F. Siddiqui, R. K. Amaravadi and J. M. Pawelek (2012). "Punctate LC3B Expression Is a Common Feature of Solid Tumors and Associated with Proliferation, Metastasis, and Poor OutcomeAutophagy in Solid Malignancies." Clinical cancer research 18(2): 370-379.
- Lee, Y. T., Y. J. Tan and C. E. Oon (2018). "Molecular targeted therapy: Treating cancer with specificity." European journal of pharmacology 834: 188-196.
- Li, T., J. Fu, Z. Zeng, D. Cohen, J. Li, Q. Chen, B. Li and X. S. Liu (2020). "TIMER2. 0 for analysis of tumor-infiltrating immune cells." Nucleic acids research 48(W1): W509-W514.
- Li, T., T. Ren, C. Huang, Y. Li, P. Yang, G. Che, L. Luo, Y. Chen, S. Peng and Y. Lin (2021). "S100A16 induces epithelial-mesenchymal transition in human PDAC cells and is a new therapeutic target for pancreatic cancer treatment that synergizes with gemcitabine." Biochemical pharmacology 189: 114396.
- Li, Y., M. Zeng, F. Zhang, F.-X. Wu and M. Li (2022). "DeepCellEss: Cell line-specific essential protein prediction with attention-based interpretable deep learning." Bioinformatics.
- Liu, C., Y. Ma, J. Zhao, R. Nussinov, Y.-C. Zhang, F. Cheng and Z.-K. Zhang (2020). "Computational network biology: data, models, and applications." Physics Reports 846: 1-66.
- Luo, W., J. Tao, L. Zheng and T. Zhang (2020). "Current epidemiology of pancreatic cancer: Challenges and opportunities." Chinese Journal of Cancer Research 32(6): 705.
- Ma, X., J. Wei, L. Zhang, D. Deng, L. Liu, X. Mei, X. He and J. Tian (2016). "miR-486-5p inhibits cell growth of papillary thyroid carcinoma by targeting fibrillin-1." Biomedicine & Pharmacotherapy 80: 220-226.
- Manser, E., T. Leung, H. Salihuddin, Z.-s. Zhao and L. Lim (1994). "A brain serine/threonine protein kinase activated by Cdc42 and Rac1." Nature 367(6458): 40-46.
- Masoudi-Sobhanzadeh, Y., A. Gholaminejad, Y. Gheisari and A. Roointan (2022). "Discovering driver nodes in chronic kidney disease-related networks using Trader as a newly developed algorithm." Computers in Biology and Medicine 148: 105892.
- McGuigan, A., P. Kelly, R. C. Turkington, C. Jones, H. G. Coleman and R. S. McCain (2018). "Pancreatic cancer: A review of clinical diagnosis, epidemiology, treatment and outcomes." World journal of gastroenterology 24(43): 4846.
- Melzer, C., R. Hass, J. von der Ohe, H. Lehnert and H. Ungefroren (2017). "The role of TGF-β and its crosstalk with RAC1/RAC1b signaling in breast and pancreas carcinoma." Cell Communication and Signaling 15(1): 1-11.
- Mira, J.-P., V. Benard, J. Groffen, L. C. Sanders and U. G. Knaus (2000). "Endogenous, hyperactive Rac3 controls proliferation of breast cancer cells by a p21-activated kinase-dependent pathway." Proceedings of the National Academy of Sciences 97(1): 185-189.
- Mizrahi, J. D., R. Surana, J. W. Valle and R. T. Shroff (2020). "Pancreatic cancer." The Lancet 395(10242): 2008-2020.
- Nickols, N. G., R. Nazarian, S. G. Zhao, V. Tan, V. Uzunangelov, Z. Xia, R. Baertsch, E. Neeman, A. C. Gao and G. V. Thomas (2019). "MEK-ERK signaling is a therapeutic target in metastatic castration resistant prostate cancer." Prostate cancer and prostatic diseases 22(4): 531-538.
- Niu, J., T. Yan, W. Guo, W. Wang, Z. Zhao, T. Ren, Y. Huang, H. Zhang, Y. Yu and X. Liang (2020). "Identification of potential therapeutic targets and immune cell infiltration characteristics in osteosarcoma using bioinformatics strategy." Frontiers in Oncology 10: 1628.
- Nowakowska, A. W. and M. Kotulska (2022). "Topological analysis as a tool for detection of abnormalities in protein–protein interaction data." Bioinformatics 38(16): 3968-3975.
- Pandey, K., N. Park, K.-S. Park, J. Hur, Y. B. Cho, M. Kang, H.-J. An, S. Kim, S. Hwang and Y. W. Moon (2020). "Combined CDK2 and CDK4/6 inhibition overcomes palbociclib resistance in breast cancer by enhancing senescence." Cancers 12(12): 3566.
- Prasad, K., S. Y. AlOmar, S. A. M. Alqahtani, M. Malik and V. Kumar (2021). "Brain disease network analysis to elucidate the neurological manifestations of COVID-19." Molecular neurobiology 58(5): 1875-1893.
- Rane, C. K. and A. Minden (2019). P21 activated kinase signaling in cancer. Seminars in cancer biology, Elsevier.
- Ren, J., J. Wang, M. Li and F. Wu (2015). "Discovering essential proteins based on PPI network and protein complex." International journal of data mining and bioinformatics 12(1): 24-43.
- Rodenburg, W. S. and J. D. van Buul (2021). "Rho GTPase signalling networks in cancer cell transendothelial migration." Vascular Biology 3(1): R77-R95.
- Roointan, A., Y. Gheisari, K. L. Hudkins and A. Gholaminejad (2021). "Non-invasive metabolic biomarkers for early diagnosis of diabetic nephropathy: Meta-analysis of profiling metabolomics studies." Nutrition, Metabolism and Cardiovascular Diseases 31(8): 2253-2272.
- Ryan, D. P., T. S. Hong and N. Bardeesy (2014). "Pancreatic adenocarcinoma." New England Journal of Medicine 371(11): 1039-1049.
- Sanchez, R. and S. A. Mackenzie (2020). "Integrative network analysis of differentially methylated and expressed genes for biomarker identification in leukemia." Scientific reports 10(1): 1-16.
- Shannon, P., A. Markiel, O. Ozier, N. S. Baliga, J. T. Wang, D. Ramage, N. Amin, B. Schwikowski and T. Ideker (2003). "Cytoscape: a software environment for integrated models of biomolecular interaction networks." Genome research 13(11): 2498-2504.
- Sharma, P., D. K. Bhattacharyya and J. K. Kalita (2016). Centrality analysis in PPI networks. 2016 International Conference on Accessibility to Digital World (ICADW), IEEE.
- Singhi, A. D. and L. D. Wood (2021). "Early detection of pancreatic cancer using DNA-based molecular approaches." Nature Reviews Gastroenterology & Hepatology 18(7): 457-468.
- Spranger, S. and T. F. Gajewski (2018). "Impact of oncogenic pathways on evasion of antitumour immune responses." Nature Reviews Cancer 18(3): 139-147.
- Szklarczyk, D., A. L. Gable, D. Lyon, A. Junge, S. Wyder, J. Huerta-Cepas, M. Simonovic, N. T. Doncheva, J. H. Morris and P. Bork (2019). "STRING v11: protein–protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets." Nucleic acids research 47(D1): D607-D613.
- Taniuchi, K., H. Nakagawa, M. Hosokawa, T. Nakamura, H. Eguchi, H. Ohigashi, O. Ishikawa, T. Katagiri and Y. Nakamura (2005). "Overexpressed P-cadherin/CDH3 promotes motility of pancreatic cancer cells by interacting with p120ctn and activating rho-family GTPases." Cancer research 65(8): 3092-3099.
- Theodosiou, T., G. Efstathiou, N. Papanikolaou, N. C. Kyrpides, P. G. Bagos, I. Iliopoulos and G. A. Pavlopoulos (2017). "NAP: The Network Analysis Profiler, a web tool for easier topological analysis and comparison of medium-scale biological networks." BMC research notes 10(1): 1-9.
- Tjomsland, V., P. Sandström, A. Spångeus, D. Messmer, J. Emilsson, U. Falkmer, S. Falkmer, K.-E. Magnusson, K. Borch and M. Larsson (2010). "Pancreatic adenocarcinoma exerts systemic effects on the peripheral blood myeloid and plasmacytoid dendritic cells: an indicator of disease severity?" BMC cancer 10(1): 1-14.
- Tomkins, J. E. and C. Manzoni (2021). "Advances in protein-protein interaction network analysis for Parkinson's disease." Neurobiology of Disease 155: 105395.
- Veenstra, T. D. (2021). "Omics in systems biology: current progress and future outlook." Proteomics 21(3-4): 2000235.
- Wang, K., Y. Zhan, N. Huynh, C. Dumesny, X. Wang, K. Asadi, D. Herrmann, P. Timpson, Y. Yang and K. Walsh (2020). "Inhibition of PAK1 suppresses pancreatic cancer by stimulation of anti-tumour immunity through down-regulation of PD-L1." Cancer letters 472: 8-18.
- Wang, Z., Y. Liu, L. Lu, L. Yang, S. Yin, Y. Wang, Z. Qi, J. Meng, R. Zang and G. Yang (2015). "Fibrillin-1, induced by Aurora-A but inhibited by BRCA2, promotes ovarian cancer metastasis." Oncotarget 6(9): 6670.
- Wood, L. D., M. I. Canto, E. M. Jaffee and D. M. Simeone (2022). "Pancreatic Cancer: Pathogenesis, Screening, Diagnosis and Treatment." Gastroenterology.
- Xie, Y.-G., Y. Yu, L.-K. Hou, X. Wang, B. Zhang and X.-C. Cao (2016). "FYN promotes breast cancer progression through epithelial-mesenchymal transition." Oncology reports 36(2): 1000-1006.
- Yan, J., S. L. Risacher, L. Shen and A. J. Saykin (2018). "Network approaches to systems biology analysis of complex disease: integrative methods for multi-omics data." Briefings in bioinformatics 19(6): 1370-1381.
- Yao, D., C. Li, M. S. R. Rajoka, Z. He, J. Huang, J. Wang and J. Zhang (2020). "P21-Activated Kinase 1: Emerging biological functions and potential therapeutic targets in Cancer." Theranostics 10(21): 9741.
- Yin, X., P. Wang, T. Yang, G. Li, X. Teng, W. Huang and H. Yu (2021). "Identification of key modules and genes associated with breast cancer prognosis using WGCNA and ceRNA network analysis." Aging (Albany NY) 13(2): 2519.
- Zhao, H., M. Yang, J. Zhao, J. Wang, Y. Zhang and Q. Zhang (2013). "High expression of LC3B is associated with progression and poor outcome in triple-negative breast cancer." Medical oncology 30(1): 1-8.
- Zhao, Z. and W. Liu (2020). "Pancreatic cancer: a review of risk factors, diagnosis, and treatment." Technology in cancer research & treatment 19: 1533033820962117.
- Zubor, P., Z. Dankova, Z. Kolkova, V. Holubekova, D. Brany, S. Mersakova, M. Samec, A. Liskova, L. Koklesova and P. Kubatka (2020). "Rho GTPases in gynecologic cancers: in-depth analysis toward the paradigm change from reactive to predictive, preventive, and personalized medical approach benefiting the patient and healthcare." Cancers 12(5): 1292.
No competing interests reported.