Introduction
Ovarian cancer (OC) is the seventh-most common malignancy in women worldwide, with the eighth-highest rate of cancer mortality, seriously threatening women’s lives.1,2 The most common subtype of OC is serous ovarian cancer. Due to its anatomic location and the silent progression of OC, approximately 80% of patients with OC are diagnosed at an advanced stage with an extremely poor prognosis.3,4 Widespread pelvic and abdominal metastasis and planting is a major risk of death with OC.5
Ovarian cancer has a tendency to metastasize to the omentum, which is rich in adipocytes.6 Omental metastasis of OC is closely associated with the metastatic progression of the tumor.7 At present, it is not clear why ovarian cancer cells preferentially metastasize to and proliferate in the omentum. Studies have revealed that OC preferentially seeds in the omental fat band, not merely because of the anatomical structure of the omentum, but also partially because a series of changes occur at the transcriptional level in adipocytes in the omentum, induced by the cancer microenvironment, that promote homing and omental metastasis of ovarian cancer cells.8 Furthermore, Nieman et al. found that lipid could directly transfer from adipocytes to ovarian cancer cells under co-culture conditions, promoting tumor growth in vitro and in vivo.9 In addition, they demonstrated that co-culture of adipocyte-ovarian cancer cells induces fat metabolism and β-oxidation in cancer cells, indicating that adipocytes serve as energy sources for cancer cells.
After migrating to the omentum, ovarian cancer cells grow much faster than primary tumor cells, suggesting that the omentum can facilitate tumor cells to migrate, colonize, invade new sites, and proliferate.9 However, research has not yet elucidated specific omentum-derived molecular factors that regulate OC progression. Gene chip technology has recently undergone rapid development, and has become widely used due to its advantages of efficiency, large-scale implementation and high throughput capabilities. At present, gene chip technology has been widely adopted to assist with the discovery of biomarkers for early diagnosis and molecular typing of ovarian tumors and for establishing prognostic models.10–12 Using bioinformatic analyses to explore the mechanisms involved in the omental metastasis of OC could facilitate the clinical management of this disease.
Herein, in order to identify genes that may be involved in the omental metastasis of OC, we used the gene expression profile GSE120196 dataset from the Gene Expression Omnibus (GEO) database to search for genes that are differentially expressed in omental adipose tissues from patients with benign gynecologic disease and patients with OC. This included performing gene ontology (GO), Kyoto Encyclopedia of Genes and Genome (KEGG), and PPI-network analyses to identify hub genes and key genes that may play significant mechanistic roles in the omental metastasis of OC and form possible therapeutic targets.
Materials and methods
Gene-Cloud of Biotechnology Information (GCBI)
GCBI (https://www.gcbi.com.cn ) is an R-based online comprehensive analysis software suite developed in Shanghai, China, which integrates genetic information, sample information, research findings, data algorithms and bioinformatics. The GCBI platform includes data from 120 million copies of genomic samples, as well as approximately 90,000 copies of tumor samples and as many as 17 million copies of genetic information.13–15 In the current study, we used the GCBI platform to identify DEGs between OC and normal control (NC; the omental tissues from four patients with benign gynecologic diseases). The DEGs were then used for GO enrichment analysis, KEGG analysis, pathway network analysis, and PPI network analysis.
Gene Expression Omnibus (GEO) DataSets
The GEO DataSets (http://www.ncbi.nlm.nih.gov/gds ) at National Centre of Biotechnology Information (NCBI) form the largest and most comprehensive public functional collection of genomic datasets.16 The key words “omental adipose tissue of ovarian cancer” were used to search GEO datasets and the gene chip dataset GSE120196. The GSE120196 dataset was obtained based on the GPL570 platform [HG-U133_Plus_2], with Affymetrix Human Genome U133 Plus 2.0 Array included for the subsequent analysis. The GSE120196 dataset contains a total of 14 samples, including 10 cancer-associated omental adipose tissues obtained from high-grade serous OC (HGSOC) patients and 4 micro-dissected omental adipose tissue samples from patients with benign gynecologic diseases.
Identification of DEGs
The GCBI laboratory provides seven function modules, including sample grouping, differential gene expression analysis, GO enrichment analysis, KEGG pathway analysis, and network analysis.17 The information from the GeneChip Sample was entered into the GCBI online laboratory (https://www.gcbi.com.cn/gclib/html/index ) to identify DEGs and conduct subsequent analyses. Thresholds of Q < 0.05,p < 0.05 and log2 (fold change) > 1.2 were applied to filter and select DEGs between OC and NC samples.
GO enrichment analysis and pathway analysis
Based on the DEGs, GO enrichment analysis and pathway analysis using the GCBI online platform were performed using thresholds of p < 0.05 and false discovery rate (FDR) < 0.05 to reveal the biological functions of DEGs.18 Pathway analysis was conducted using KEGG enrichment analysis with, the p-value for identifying metabolic processes set at p < 0.05. Furthermore, pathway network analysis was constructed to identify pathway connections and core networks using the GCBI platform. By assessing the upstream and downstream relationship between pathways, a deeper and broader understanding of signaling pathways can be established.
Protein-protein interaction (PPI) network and hub gene selection
Search Tool for the Retrieval of Interacting Genes (STRING) is a database of known and predicted protein-protein interactions. In this study, protein interaction networks were constructed using STRING version 11.0 online software (www.string-db.org/ ).19 The top 100 DEGs were then mapped to STRING, and the interactions with a combined score of > 0.4 for PPI pairs were selected for construction of a PPI network using Cytoscape version 3.8.0 software. Hub genes and subnetworks were identified through network analysis using the cytoHubba plugin from Cytoscape v3.8.0.20 The betweenness centrality values of the top 10 nodes in the networks were computed and were generated. Genes in this module were regarded as hub genes.
Statistical analysis
Kaplan-Meier survival analysis
Survival was analyzed using Kaplan-Meier survival analysis (https://kmplot.com/analysis/ ). Prognostic analysis was performed for progression free survival (PFS) or overall survival (OS). The following retrieval indexes were used: Gene symbol ‘use multiple genes’ field: enter top 10 hub gene name; split patients by: Upper Tertile; Stage: 3+4 stage; Survival: PFS or OS respectively. A p-value <0.01 was considered statistically significant.
Oncomine analysis
The Oncomine database (http://www.oncomine.org ) is a cancer gene chip database and integrated data-mining platform. It includes gene information covering 33 different cancers and can be used to analyze the differential expression of genes between cancer and adjacent normal tissue. In this study, the mRNA expression levels of COL1A1, VCAN, and MARS in ovarian cancer tissues and normal tissues were acquired. The settings were as follows: Gene: COL1A1, VCAN, and MARS, respectively; Analysis Type: Cancer vs Normal analysis; Cancer type: Ovarian cancer; Sample type: Clinical specimen; Data type: mRNA; Threshold (p-value < 0.001, fold change > 1.5, gene rank = top 10%).
Results
Identification of differentially expressed genes (DEGs)
To identify DEGs in omental adipose tissues from ovarian cancer patients, the transcription profile dataset GSE120196 was obtained from the NCBI GEO database, which includes information from 10 patients with HGSOC and 4 patients with benign gynecologic diseases as controls. Thresholds were set at Q < 0.05, p < 0.05 and log2 (fold change) > 1.2 for differential expression analysis. A total of 429 DEGs were identified, which included 301 genes that were significantly up-regulated and 128 genes that were significantly down-regulated. These results were visualized as a heat map and a volcano plot (Fig. 1). The top 12 genes displaying the most significant changes in expression were unnamed, glycoprotein M6A (GPM6A), serpin peptidase inhibitor, clade E, member 2 (SERPINE2), collagen type I alpha 1 (COL1A1), transmembrane 4L six family member 1(TM4SF1) , chromosome 2 open reading frame 40 (C2orf40), unnamed, solute carrier family 24 member 3 (SLC24A3), eukaryotic translation initiation factor 1(EIF1), glycine-N-acyltransferase (GLYAT), UDP-glucose glycoprotein glucosyltransferase 2 (UGGT2), and KN motif and ankyrin repeat domains 4 (KANK4) (Table 1).
Table 1The top 12 genes with the most significant differential expression
| Gene symbol | Gene description | p-value | Gene feature | Rank |
|---|
| Unnamed | – | 1.90E−05 | down | 1 |
| GPM6A | glycoprotein M6A | 2.00E−05 | down | 2 |
| SERPINE2 | serpin peptidase inhibitor, clade E, member 2 | 2.10E−05 | up | 3 |
| COL1A1 | collagen, type I, alpha 1 | 2.20E−05 | up | 4 |
| TM4SF1 | transmembrane 4 L six family member 1 | 2.20E−05 | down | 5 |
| C2orf40 | chromosome 2 open reading frame 40 | 2.40E−05 | down | 6 |
| Unnamed | – | 2.40E−05 | down | 7 |
| SLC24A3 | solute carrier family 24, member 3 | 2.50E−05 | down | 8 |
| EIF1 | eukaryotic translation initiation factor 1 | 2.60E−05 | down | 9 |
| GLYAT | glycine-N-acyltransferase | 2.80E−05 | down | 10 |
| UGGT2 | UDP-glucose glycoprotein glucosyltransferase 2 | 2.80E−05 | up | 11 |
| KANK4 | KN motif and ankyrin repeat domains 4 | 3.00E−05 | down | 12 |
GO enrichment analysis of DEGs
To better understand the biological functions of these DEGs, GO enrichment analysis was performed using the platform of Gene-Cloud of Biotechnology Information (GCBI) for all DEGs. GO analysis was conducted based on Fisher exact test. Using statistical difference thresholds of p < 0.05 and FDR < 0.05, we identified 145 sequences potentially involved in biological processes. The most common biological processes identified were negative regulation of cell proliferation, extracellular matrix organization, positive regulation of cell proliferation, collagen fibril organization, negative regulation of apoptotic process, collagen catabolic process, extracellular matrix disassembly, cell adhesion, skeletal system development, positive regulation of ERK1 and ERK2 cascade (Table 2 and Fig. 2a).
Table 2The top 20 significant GO terms for biological processes. FDR indicates false discovery rate
| BP name | Enrichment score | p-value | FDR | Rank |
|---|
| Negative regulation of cell proliferation | 7.94 | 8.82E−13 | 6.29E−10 | 1 |
| Extracellular matrix organization | 10.96 | 9.14E−13 | 6.29E−10 | 2 |
| Positive regulation of cell proliferation | 7.25 | 1.51E−12 | 6.90E−10 | 3 |
| Collagen fibril organization | 31.25 | 2.04E−11 | 7.02E−09 | 4 |
| Negative regulation of apoptotic process | 5.87 | 2.52E−10 | 6.89E−08 | 5 |
| Collagen catabolic process | 18.81 | 3.01E−10 | 6.89E−08 | 6 |
| Extracellular matrix disassembly | 17.14 | 7.72E−10 | 1.52E−07 | 7 |
| Cell adhesion | 5.67 | 3.53E−09 | 6.07E−07 | 8 |
| Skeletal system development | 11.73 | 6.26E−09 | 9.56E−07 | 9 |
| Positive regulation of ERK1 and ERK2 cascade | 14.01 | 3.62E−08 | 4.98E−06 | 10 |
| Transforming growth factor beta receptor signaling pathway | 11.19 | 5.16E−08 | 6.45E−06 | 11 |
| Positive regulation of transcription, DNA-dependent | 4.96 | 7.21E−08 | 8.26E−06 | 12 |
| Negative regulation of cell growth | 10.42 | 4.81E−07 | 5.08E−05 | 13 |
| Negative regulation of platelet activation | 60.18 | 7.15E−07 | 7.02E−05 | 14 |
| Blood vessel development | 20.83 | 8.24E−07 | 7.55E−05 | 15 |
| Positive regulation of MAPK cascade | 14.81 | 9.73E−07 | 8.36E−05 | 16 |
| Positive regulation of protein kinase B signaling cascade | 13.94 | 1.48E−06 | 1.20E−04 | 17 |
| Chondroitin sulfate biosynthetic process | 27.08 | 2.01E−06 | 1.53E−04 | 18 |
| Cell-cell signaling | 6.15 | 4.40E−06 | 3.19E−04 | 19 |
| Platelet activation | 6.64 | 6.54E−06 | 4.50E−04 | 20 |
KEGG pathway enrichment analysis of DEGs
KEGG is a powerful tool for the analysis of metabolic pathways and networks. The GCBI platform-based KEGG pathway enrichment analysis of DEGs showed that DEGs were enriched in 43 pathways with a statistical significance of p < 0.05. The enriched pathways were mainly in categories such as “PI3K-Akt signaling” (p = 3.17E-10), “Cytokine-cytokine receptor interaction” (p = 3.09E−08), “Focal adhesion” (p = 9.09E−07), “ECM-receptor interaction” (p = 7.92E−06), “Protein digestion and absorption” (p = 8.55E−06), “Jak-STAT signaling” (p = 5.01E−05), “Malaria” (p = 6.23E−05), “Pathways in cancer” (p = 7.38E−05), “Protein processing in endoplasmic reticulum” (p = 7.44E−05), and “Transcriptional mis-regulation in cancer” (p = 1.26E−04). The top 20 enriched pathway terms are displayed in Figure 2b.
Pathway network analysis of significant pathways
Pathway network analysis was constructed based on the interaction relationships provided by the KEGG database. From this network, we analyzed signal transduction of the significant upstream regulatory signaling pathways and downstream effectors in order to better understand the nature of the differences between the samples. A total of 17 pathways were identified, of which 11 were up-regulated. Six differentially regulated pathways were significantly enriched. We identified 31 relationships between these pathways (Fig. 3). As shown in Table 3, the main signaling pathways implicated in cancer-associated omental tissue were: pathways in cancer, focal adhesion, Wnt signaling, TGF-beta signaling, and cytokine-cytokine receptor interaction. Pathways in cancer was the key node in this network with the most extensive interactions with the remaining pathways.
Table 3The 10 highest scoring pathway in pathway network analysis
| Pathway name | Outdegree | Indegree | Degree | Pathway feature |
|---|
| Pathways in cancer | 14 | 0 | 14 | up|down |
| Focal adhesion | 4 | 4 | 8 | up |
| Wnt signaling pathway | 2 | 3 | 5 | up|down |
| TGF-beta signaling pathway | 0 | 5 | 5 | up |
| Cytokine-cytokine receptor interaction | 0 | 5 | 5 | down|up |
| ECM-receptor interaction | 1 | 2 | 3 | up |
| Jak-STAT signaling pathway | 1 | 2 | 3 | down|up |
| Regulation of actin cytoskeleton | 1 | 2 | 3 | down|up |
| Pancreatic cancer | 2 | 1 | 3 | up |
| Basal cell carcinoma | 2 | 1 | 3 | up |
Construction of the PPI network and identification of hub genes
PPI network analysis constructs a network of protein interactions as well as identifying hub genes. The PPI network of the top 100 DEGs obtained from the GSE120196 dataset was constructed using the STRING database and Cytoscape software. This revealed a PPI network containing 43 hub nodes and 62 edges (Fig. 4a). The top 10 of these 43 genes were screened as hub genes using the betweenness method of the cytoHubba plugin. These genes included TP53, COL1A1, HSPD1, SERPINE2, CSF3, MARS, F2R, VCAN, COL1A2, and THBD (Table 4). The expanded subnetwork of the top 10 hub genes is shown in Figure 4b.
Table 4The top 10 hub genes were identified by cytoHubba
| Gene symbol | Gene feature | Gene description | Betweenness |
|---|
| TP53 | up | tumor protein p53 | 864.2 |
| COL1A1 | up | collagen, type I, alpha 1 | 368.23 |
| HSPD1 | down | heat shock 60kDa protein 1 (chaperonin) | 186 |
| SERPINE2 | up | serpin peptidase inhibitor, clade E (nexin, plasminogen activator inhibitor type 1), member 2 | 170.2 |
| CSF3 | down | colony stimulating factor 3 (granulocyte) | 144 |
| MARS | up | methionyl-tRNA synthetase | 128 |
| F2R | up | coagulation factor II (thrombin) receptor | 105.33 |
| VCAN | up | versican | 76.47 |
| COL1A2 | up | collagen, type I, alpha 2 | 72.7 |
| THBD | down | thrombomodulin | 69.67 |
Kaplan-Meier analysis of hub genes
In the above analyses, we studied gene networks derived from OC with greater omentum metastases. To investigate the association between gene networks and OC further, we performed Kaplan-Meier survival analyses to investigate the association of ten hub genes with PFS and OS of advanced-stage ovarian cancer (FIGO Stage III–IV). We found that patients with higher expression levels of COL1A1 or VCAN had significantly lower PFS rates, and patients with higher expression levels of MARS had longer PFS rates (Fig. 5a), with these differences being statistically significant (p = 0.0022, p = 0.0025, and p = 0.00017, respectively). Moreover, survival analysis showed that only VCAN expression was negatively correlated with overall survival (Fig. 5b, p = 0.0031). The expression levels of the remaining genes were not significantly associated with survival.
Validation of gene expression based on the Oncomine database
To further investigate mRNA expression levels of COL1A1, VCAN and MARS in ovarian cancer, our study used the Oncomine database. Results showed that COL1A1, VCAN and MARS mRNA expression levels were significantly higher in tumor samples than samples from non-cancerous tissues (see Figs. 6a, 7a, 8a). To further understand the expression of these genes in human ovarian cancer, their expression level in normal ovary vs. ovarian cancer tissues were compared. COL1A1, VCAN and MARS mRNA levels were significantly higher in ovarian cancer tissues than those in the adjacent normal tissues (Fig. 6b, p = 1.99E−12; Fig. 7b, p = 2.88E−4; Fig. 8b, p = 1.02E−4), and were each highly expressed in different histological types of human ovarian cancer (all p < 0.001; Fig. 6c, Fig. 7c and Fig. 8c).
In summary, survival analysis identified three up-regulated hub genes that were associated with survival. Increased COL1A1 and VCAN expression correlated with poorer survival, while in contrast higher expression of MARS was associated with longer PFS. Since COL1A1 and VCAN expression levels were significantly correlated with the prognosis of advanced OC patients, they may be useful as novel targets for diagnosis and therapy in OC patients with omental metastasis.
Discussion
Ovarian cancer metastasis is a complex process that involves numerous influential factors and multiple steps. The metastatic routes of OC principally consist of direct spread, intraperitoneal implantation metastasis and lymphatic metastasis. Hematogenous metastasis only occurs rarely.9,21 Exfoliated tumor cells from advanced OC are widely disseminated in the abdominal-pelvic cavity, and the greater omentum is the most common metastatic site of ovarian cancer.7 The greater omentum is an important immune organ in the abdominopelvic cavity, which covers the surface of the viscera and is richly supplied with blood vessels, lymphatic drainage, adipose tissue, and nerve fibers. It also is known to have anti-tumor properties. However, in advanced ovarian cancer the greater omentum is often contracted into a pie shape due to extensive seeding of tumor cells. Dynamic observation of animal models of ovarian tumors conducted at the University of Chicago showed that ovarian cancer cells infiltrate the greater omentum within 20 minutes of entering the abdominal cavity.22 Therefore, the study of omental metastasis in ovarian cancer has become an important focus of research in recent years. Nowicka isolated pluripotential adipose-derived stem cells from ovarian cancer patients’ greater omentum, indicating that adipose-derived stem cells could promote the proliferation and migration of tumor cells and contribute to their resistance to chemotherapy.23 Additionally, previous studies have shown that ovarian cancer cells can induce lipolysis in the omentum, which can provide energy for the invasion and metastasis of tumor cells.9 Despite progress in studying on omental metastasis of OC, the underlying mechanisms are still not fully elucidated.
Gene chip technology makes it possible to explore specific mechanisms of ovarian cancer occurrence at the genomic level and then focus more precisely on the identified gene targets. In this study, we sought to explore and analyze the gene chip detection data of RNA samples from cancer-associated omental tissues derived from ten patients with OC and benign control tissues derived from four patients in order to identify differentially expressed genes related to the omental metastasis of OC. Our findings show significant differences in the gene expression profiles of the two sets of tissues. Among more than 50,000 genes detected, a total of 429 genes are differentially expressed, including 301 up-regulated genes and 128 down-regulated genes, potentially identifying genes affecting tumor development. We further screened the top 12 genes identified by these analyses (unnamed, GPM6A, SERPINE2, COL1A1, TM4SF1, C2orf40, unnamed, SLC24A3, EIF1, GLYAT, UGGT2, and KANK4). Of special note, expression of GPM6A was the lowest.
GO analysis revealed that differentially expressed genes have functions that are mainly related to cell proliferation, extracellular matrix, collagen, and cell adhesion. The PI3K-Akt signaling pathway, cytokine-cytokine receptor interaction, focal adhesion, ECM-receptor interaction, protein digestion and absorption, Jak-STAT signaling, and other pathways in cancer were also identified by KEGG pathway enrichment analysis. Prior research has shown that the progression of malignant tumors is often accompanied by changes in the construction of the extracellular matrix and expression of cell surface receptors,24,25 consistent with our findings in cancer-associated omental tissues from the patients with advanced stage, high-grade serous ovarian cancer (HGSOC) in the current study. In addition, changes in collagen content within tumor cells and the tumor microenvironment play an essential role in the invasion and metastasis of tumor cells by affecting tumor metabolism, gene expression, angiogenesis, and epithelial-mesenchymal transition.26–30
In the current study, we found that genes that were significantly differentially expressed were involved in 43 pathways, including PI3K-Akt signaling, cytokine-cytokine receptor interaction, focal adhesion, and ECM-receptor interaction. The activation of PI3K-AKT pathway is important for cancer cell proliferation, migration, invasion, and chemoresistance, including in ovarian cancer cells.31,32 In the PPI network analysis, a total of ten hub genes formed the center of the network: TP53, COL1A1, HSPD1, SERPINE2, CSF3, MARS, F2R, VCAN, COL1A2 and THBD. With the exception of HSPD1, CSF3 and THBD, each of these genes were up-regulated in metastatic ovarian cancer-associated samples compared to samples from patients with benign gynecologic diseases.
The present study implicated several DEGs in the development and progression of ovarian cancer, including COL1A1, VCAN, and MARS. In addition, examination of the association between expression levels of ten hub genes and patients’ PFS and OS demonstrated that higher expression levels of VCAN were associated with worse PFS and OS, higher expression levels of COL1A1 were associated with reduced PFS, and that higher expression of MARS correlated with increased PFS. These findings on MARS are not consistent with previous studies in which MARS expression is upregulated and positively correlated with poor prognosis in various cancers, such as breast cancer and lung cancer. Taken together, we conclude that MARS may not be a regulator for advanced ovarian cancer. COL1A1 and VCAN, which were identified as key genes associated with OC survival in the current study, could be useful as diagnostic and prognostic biomarkers.
Studies have suggested that aberrant expression of COL1A1 is closely related to tumorigenesis in several malignancies.33–36COL1A1 expression can induce the epithelial-mesenchymal transition, thus promoting invasion and metastasis of tumor cells.37 Liu et al. confirmed that COL1A1 is a potential therapeutic target for breast cancer as it mediates the metastatic process.38 Recent studies have suggested that COL1A1 may be a biomarker for early detection of gastric cancer.39 Moreover, up-regulation of COL1A1 may contribute to cisplatin resistance in ovarian cancer cells.40 Our results show that COL1A1 expression was significantly up-regulated in OC-associated omental tissues. In addition, patients with higher COL1A1 expression had worse PFS.
VCAN belongs to the family of large aggregating chondroitin sulfate proteoglycans and is mainly found in the extracellular matrix.41VCAN expression levels have been shown to correlate with tumor progression, and are a poor prognostic indicator in stage II–III colon cancer patients.42 Kulbe et al. have suggested that VCAN has the potential to be used as a biomarker for ovarian cancer,43 as stromal VCAN is a biomarker with a poor prognosis.44 Our study showed that high VCAN expression is associated with worse PFS and OS in patients with advanced OC, suggesting that it may be implicated in the omental metastasis of OC.
Future directions
Herein, we identified two key genes, COL1A1 and VCAN, which may play a key role in the omental metastasis of OC. These genes may be useful as new prognostic biomarkers for advanced OC. Further studies are needed to validate the roles played by COL1A1 and VCAN genes in advanced OC and their role in the progression of early stages of OC.
Conclusions
This study used comprehensive bioinformatics to identify two genes (COL1A1 and VCAN), which may serve key roles in the tumorigenesis of OC with omental metastasis (Fig. 9). These key genes could serve as novel prognostic biomarkers of advanced OC. However, a limitation of the current study is the lack of in vivo and in vitro experiments. Further experiments are required to confirm the present findings, and explore the specific mechanisms and the exact functions of these genes in advanced-stage OC.
Abbreviations
- OC:
ovarian cancer
- HGSOC:
high-grade serous ovarian cancer
- DEGs:
differentially expressed genes
- GO:
gene ontology
- KEGG:
Kyoto Encyclopedia of Genes and Genomes
- PPI network:
protein-protein interaction network
- GCBI:
Gene-Cloud of Biotechnology Information
- GEO:
Gene Expression Omnibus
- NCBI:
National Centre of Biotechnology Information
- GPM6A:
glycoprotein M6A
- SERPINE2:
serpin peptidase inhibitor, clade E, member 2
- COL1A1:
collagen type I alpha 1
- TM4SF1:
transmembrane 4L six family member 1
- TP53:
tumor protein p53
- HSPD1:
heat shock protein family D (Hsp60) member 1
- CSF3:
colony stimulating factor 3
- MARS:
methionyl-tRNA synthetase
- F2R:
coagulation factor II (thrombin) receptor
- VCAN:
versican
- C2orf40:
chromosome 2 open reading frame 40
- SLC24A3:
solute carrier family 24 member 3
- EIF1:
eukaryotic translation initiation factor 1
- GLYAT:
glycine-N-acyltransferase
- UGGT2:
UDP-glucose glycoprotein glucosyltransferase 2
- KANK4:
KN motif and ankyrin repeat domains 4
- STRING:
Search Tool for the Retrieval of Interacting Genes
Declarations
Acknowledgement
We are grateful to researchers who have deposited their data in the GEO database. The authors would like to acknowledge the technical assistance of the GCBI online analysis platform.
Data sharing statement
The datasets generated and/or analyzed in the present study are available in the GEO (http://www.ncbi.nlm.nih.gov/gds) and the Oncomine (http://www.oncomine.org) repositories.
Funding
This work was supported by grants from Hainan health science and education project [Item No. 21A200020] and Hainan health science and education project [Item No. 20A200415].
Conflict of interest
The authors declare no competing interests.
Authors’ contributions
MZ and LW conceived and designed the project; MZ and JT performed the analysis and constructed the model; LS, HW and XZ checked the associated database and classified raw data; MZ and JT wrote and revised the manuscript. All of the authors read and approved the final manuscript.