Identification of key genes for HNSCC from public databases using bioinformatics analysis

Abstract Background The cause and underlying molecular mechanisms of head and neck squamous cell carcinoma (HNSCC) are unclear. Our study aims to identify the key genes associated with HNSCC and reveal potential biomarkers. Methods In this study, the expression profile dataset GSE83519 of the Gene Expression Omnibus database and the RNA sequencing dataset of HNSCC of The Cancer Genome Atlas were included for analysis. Sixteen differentially expressed genes were screened from these two datasets using R software. Gene Expression Profiling Interactive Analysis 2 (GEPIA2) was then adopted for survival analysis, and finally, three key genes related to the overall survival of HNSCC patients were identified. Furthermore, we verified these three genes using the Oncomine database and from real-time PCR and immunohistochemistry results from HNSCC tissues. Results The expression data of 44 samples from GSE83519 and 545 samples from TCGA-HNSC were collected. Using bioinformatics, the two databases were integrated, and 16 DEGs were screened out. Gene Ontology (GO) enrichment analysis showed that the biological functions of DEGs focused primarily on the apical plasma membrane and regulation of anoikis. Kyoto Encyclopedia of Genes and Genomes (KEGG) signalling pathway analysis showed that these DEGs were mainly involved in drug metabolism-cytochrome P450 and serotonergic synapses. Survival analysis identified three key genes, CEACAM5, CEACAM6 and CLCA4, that were closely related to HNSCC prognosis. The Oncomine database, qRT–PCR and IHC verified that all 3 key genes were downregulated in most HNSCC tissues compared to adjacent normal tissues. Conclusions This study indicates that integrated bioinformatics analyses play an important role in screening for differentially expressed genes and pathways in HNSCC, helping us better understand the biomarkers and molecular mechanism of HNSCC.

Download Full-text

AB0005 IDENTIFICATION OF KEY GENES AND PATHWAYS FOR PSORIASIS BASED ON GEO DATABASES BY BIOINFORMATICS ANALYSIS

Annals of the Rheumatic Diseases ◽

10.1136/annrheumdis-2021-eular.1773 ◽

2021 ◽

Vol 80 (Suppl 1) ◽

pp. 1037.2-1038

Author(s):

X. Sun ◽

S. X. Zhang ◽

S. Song ◽

T. Kong ◽

C. Zheng ◽

...

Keyword(s):

Gene Expression ◽

Signaling Pathways ◽

Differentially Expressed Genes ◽

Enrichment Analysis ◽

Differentially Expressed ◽

Shanxi Province ◽

Receptor Interaction ◽

Term Therapy ◽

Pathway Enrichment Analysis ◽

Key Genes

Background:Psoriasis is an immune-mediated, genetic disease manifesting in the skin or joints or both, and also has a strong genetic predisposition and autoimmune pathogenic traits1. The hallmark of psoriasis is sustained inflammation that leads to uncontrolled keratinocyte proliferation and dysfunctional differentiation. And it’s also a chronic relapsing disease, which often necessitates a long-term therapy2.Objectives:To investigate the molecular mechanisms of psoriasis and find the potential gene targets for diagnosis and treating psoriasis.Methods:Total 334 gene expression data of patients with psoriasis research (GSE13355 GSE14905 and GSE30999) were obtained from the Gene Expression Omnibus database. After data preprocessing and screening of differentially expressed genes (DEGs) by R software. Online toll Metascape3 was used to analyze Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of DEGs. Interactions of proteins encoded by DEGs were discovered by Protein-protein interaction network (PPI) using STRING online software. Cytoscape software was utilized to visualize PPI and the degree of each DEGs was obtained by analyzing the topological structure of the PPI network.Results:A total of 611 DEGs were found to be differentially expressed in psoriasis. GO analysis revealed that up-regulated DEGs were mostly associated with defense and response to external stimulus while down-regulated DEGs were mostly associated with metabolism and synthesis of lipids. KEGG enrichment analysis suggested they were mainly enriched in IL-17 signaling, Toll-like receptor signaling and PPAR signaling pathways, Cytokine-cytokine receptor interaction and lipid metabolism. In addition, top 9 key genes (CXCL10, OASL, IFIT1, IFIT3, RSAD2, MX1, OAS1, IFI44 and OAS2) were identified through Cytoscape.Conclusion:DEGs of psoriasis may play an essential role in disease development and may be potential pathogeneses of psoriasis.References:[1]Boehncke WH, Schon MP. Psoriasis. Lancet 2015;386(9997):983-94. doi: 10.1016/S0140-6736(14)61909-7 [published Online First: 2015/05/31].[2]Zhang YJ, Sun YZ, Gao XH, et al. Integrated bioinformatic analysis of differentially expressed genes and signaling pathways in plaque psoriasis. Mol Med Rep 2019;20(1):225-35. doi: 10.3892/mmr.2019.10241 [published Online First: 2019/05/23].[3]Zhou Y, Zhou B, Pache L, et al. Metascape provides a biologist-oriented resource for the analysis of systems-level datasets. Nat Commun 2019;10(1):1523. doi: 10.1038/s41467-019-09234-6 [published Online First: 2019/04/05].Acknowledgements:This project was supported by National Science Foundation of China (82001740), Open Fund from the Key Laboratory of Cellular Physiology (Shanxi Medical University) (KLCP2019) and Innovation Plan for Postgraduate Education in Shanxi Province (2020BY078).Disclosure of Interests:None declared

Download Full-text

Key Genes and Prognostic Analysis in HER2+ Breast Cancer

Technology in Cancer Research & Treatment ◽

10.1177/1533033820983298 ◽

2021 ◽

Vol 20 ◽

pp. 153303382098329

Author(s):

Yujie Weng ◽

Wei Liang ◽

Yucheng Ji ◽

Zhongxian Li ◽

Rong Jia ◽

...

Keyword(s):

Breast Cancer ◽

Gene Expression ◽

Differentially Expressed Genes ◽

Protein Interaction ◽

Differentially Expressed ◽

Protein Kinase Activity ◽

Protein Protein Interaction ◽

Risk Of Death ◽

Her2 Breast Cancer ◽

Key Genes

Human epidermal growth factor 2 (HER2)+ breast cancer is considered the most dangerous type of breast cancers. Herein, we used bioinformatics methods to identify potential key genes in HER2+ breast cancer to enable its diagnosis, treatment, and prognosis prediction. Datasets of HER2+ breast cancer and normal tissue samples retrieved from Gene Expression Omnibus and The Cancer Genome Atlas databases were subjected to analysis for differentially expressed genes using R software. The identified differentially expressed genes were subjected to gene ontology and Kyoto Encyclopedia of Genes and Genomes pathway enrichment analyses followed by construction of protein-protein interaction networks using the STRING database to identify key genes. The genes were further validated via survival and differential gene expression analyses. We identified 97 upregulated and 106 downregulated genes that were primarily associated with processes such as mitosis, protein kinase activity, cell cycle, and the p53 signaling pathway. Visualization of the protein-protein interaction network identified 10 key genes ( CCNA2, CDK1, CDC20, CCNB1, DLGAP5, AURKA, BUB1B, RRM2, TPX2, and MAD2L1), all of which were upregulated. Survival analysis using PROGgeneV2 showed that CDC20, CCNA2, DLGAP5, RRM2, and TPX2 are prognosis-related key genes in HER2+ breast cancer. A nomogram showed that high expression of RRM2, DLGAP5, and TPX2 was positively associated with the risk of death. TPX2, which has not previously been reported in HER2+ breast cancer, was associated with breast cancer development, progression, and prognosis and is therefore a potential key gene. It is hoped that this study can provide a new method for the diagnosis and treatment of HER2 + breast cancer.

Download Full-text

Diagnostic model of combined ceRNA and DNA methylation related genes in esophageal carcinoma

PeerJ ◽

10.7717/peerj.8831 ◽

2020 ◽

Vol 8 ◽

pp. e8831 ◽

Cited By ~ 1

Author(s):

Xiaojiao Guan ◽

Yao Yao ◽

Guangyao Bao ◽

Yue Wang ◽

Aimeng Zhang ◽

...

Keyword(s):

Gene Expression ◽

Esophageal Cancer ◽

Esophageal Carcinoma ◽

Differentially Expressed Genes ◽

Enrichment Analysis ◽

Functional Enrichment ◽

Differentially Expressed ◽

Diagnostic Model ◽

Link Type ◽

Key Genes

Esophageal cancer is a common malignant tumor in the world, and the aim of this study was to screen key genes related to the development of esophageal cancer using a variety of bioinformatics analysis tools and analyze their biological functions. The data of esophageal squamous cell carcinoma from the Gene Expression Omnibus (GEO) were selected as the research object, processed and analyzed to screen differentially expressed microRNAs (miRNAs) and differential methylation genes. The competing endogenous RNAs (ceRNAs) interaction network of differentially expressed genes was constructed by bioinformatics tools DAVID, String, and Cytoscape. Biofunctional enrichment analysis was performed using Gene Ontology (GO) and the Kyoto Encyclopedia of Genes and Genomes (KEGG). The expression of the screened genes and the survival of the patients were verified. By analyzing GSE59973 and GSE114110, we found three down-regulated and nine up-regulated miRNAs. The gene expression matrix of GSE120356 was calculated by Pearson correlation coefficient, and the 11696 pairs of ceRNA relation were determined. In the ceRNA network, 643 lncRNAs and 147 mRNAs showed methylation difference. Functional enrichment analysis showed that these differentially expressed genes were mainly concentrated in the FoxO signaling pathway and were involved in the corresponding cascade of calcineurin. By analyzing the clinical data in The Cancer Genome Atlas (TCGA) database, it was found that four lncRNAs had an important impact on the survival and prognosis of esophageal carcinoma patients. QRT-PCR was also conducted to identify the expression of the key lncRNAs (RNF217-AS1, HCP5, ZFPM2-AS1 and HCG22) in ESCC samples. The selected key genes can provide theoretical guidance for further research on the molecular mechanism of esophageal carcinoma and the screening of molecular markers.

Download Full-text

Insight into Molecular Profile Changes after Skeletal Muscle Contusion Using Microarray and Bioinformatics Analyses

10.21203/rs.3.rs-85830/v1 ◽

2020 ◽

Author(s):

Na Li ◽

Ru-feng Bai ◽

Chun Li ◽

Li-hong Dang ◽

Qiu-xiang Du ◽

...

Keyword(s):

Gene Expression ◽

Network Analysis ◽

Differentially Expressed Genes ◽

Molecular Mechanisms ◽

Expression Profiles ◽

Healing Process ◽

Differentially Expressed ◽

Molecular Profile ◽

Wound Age ◽

Functional Modules

Abstract Background: Muscle trauma frequently occurs in daily life. However, the molecular mechanisms of muscle healing, which partly depend on the extent of the damage, are not well understood. This study aimed to investigate gene expression profiles following mild and severe muscle contusion, and to provide more information about the molecular mechanisms underlying the repair process.Methods: A total of 33 rats were divided randomly into control (n = 3), mild contusion (n = 15), and severe contusion (n = 15) groups; the contusion groups were further divided into five subgroups (1, 3, 24, 48, and 168 h post-injury; n = 3 per subgroup). Then full genome microarray of RNA isolated from muscle tissue was performed to access the gene expression changes during healing process.Results: A total of 2,844 and 2,298 differentially expressed genes were identified in the mild and severe contusion groups, respectively. The analysis of the overlapping differentially expressed genes showed that there are common mechanisms of transcriptomic repair of mild and severe contusion within 48 h post-contusion. This was supported by the results of principal component analysis, hierarchical clustering, and weighted gene co‐expression network analysis of the 1,620 coexpressed genes in mildly and severely contused muscle. From these analyses, we discovered that the gene profiles in functional modules and temporal clusters were similar between the mild and severe contusion groups; moreover, the genes showed time-dependent patterns of expression, which allowed us to identify useful markers of wound age. We then performed an analysis of the functions of genes (including Gene Ontology and Kyoto Encyclopedia of Genes and Genomes pathway annotation, and protein–protein interaction network analysis) in the functional modules and temporal clusters, and the hub genes in each module–cluster pair were identified. Interestingly, we found that genes downregulated within 24−48 h of the healing process were largely associated with metabolic processes, especially oxidative phosphorylation of reduced nicotinamide adenine dinucleotide phosphate, which has been rarely reported. Conclusions: These results improve our understanding of the molecular mechanisms underlying muscle repair, and provide a basis for further studies of wound age estimation.

Download Full-text

Identification of Hypoxia-Related Differentially Expressed Genes and Construction of the Clinical Prognostic Predictor in Hepatocellular Carcinoma by Bioinformatic Analysis

BioMed Research International ◽

10.1155/2021/7928051 ◽

2021 ◽

Vol 2021 ◽

pp. 1-15

Author(s):

Hao Guo ◽

Jing Zhou ◽

Yanjun Zhang ◽

Zhi Wang ◽

Likun Liu ◽

...

Keyword(s):

Hepatocellular Carcinoma ◽

Regression Model ◽

Molecular Mechanisms ◽

Cox Regression ◽

The Cancer Genome Atlas ◽

Functional Enrichment ◽

Differentially Expressed ◽

Clinical Prognosis ◽

Prognostic Predictor ◽

Key Genes

Background. Hypoxia closely relates to malignant progression and appears to be prognostic for outcome in hepatocellular carcinoma (HCC). Our research is aimed at mining the hypoxic-related genes (HRGs) and constructing a prognostic predictor (PP) model on clinical prognosis in HCC patients. Methods. RNA-sequencing data about HRGs and clinical data of patients with HCC were obtained from The Cancer Genome Atlas (TCGA) database portal. Differentially expressed HRGs between HCC and para-carcinoma tissue samples were obtained by applying the Wilcox analysis in R statistical software. Gene Ontology and Kyoto Encyclopedia of Genes and Genomes were used for gene functional enrichment analyses. Then, the patients who were asked to follow up for at least one month were enrolled in the following study. Cox proportional risk regression model was applied to obtain key HRGs which related to overall survival (OS) in HCC. PP was constructed and defined, and the accuracy of PP was validated by constructing the signature in a training set and validation set. Connectivity map (CMap) was used to find potential drugs, and gene set cancer analysis (GSCA) was also performed to explore the underlying molecular mechanisms. Results. Thirty-seven differentially expressed HRGs were obtained. It contained 28 upregulated and 9 downregulated genes. After the univariate Cox regression model analysis, we obtained 27 prognosis-related HRGs. Of these, 25 genes were risk factors for cancer, and 2 genes were protective factors. The PP was composed by 12 key genes (HDLBP, SAP30, PFKP, DPYSL4, SLC2A1, HMOX1, PGK1, ERO1A, LDHA, ENO2, SLC6A6, and TPI1). GSCA results showed the overall activity of these 12 key genes in 10 cancer-related pathways. Besides, CMap identified deferoxamine, crotamiton, talampicillin, and lycorine might have effects with HCC. Conclusions. This study firstly reported 12 prognostic HRGs and constructed the model of the PP. This comprehensive research of multiple databases helps us gain insight into the biological properties of HCC and provides deferoxamine, crotamiton, talampicillin, and lycorine as potential drugs to fight against HCC.

Download Full-text

A Comprehensive Data Analysis of Differentially Regulated Genes in Melanoma

10.21203/rs.3.rs-19035/v1 ◽

2020 ◽

Author(s):

Yanjie Han ◽

Xinxin Li ◽

Jiliang Yan ◽

Chunyan Ma ◽

Xin Wang ◽

...

Keyword(s):

Differentially Expressed Genes ◽

Molecular Mechanisms ◽

Expression Profiles ◽

Growth Factor Receptor ◽

Principal Component ◽

Distant Metastases ◽

The Cancer Genome Atlas ◽

Differentially Expressed ◽

Hub Genes ◽

Ppi Networks

Abstract Background: Melanoma is the most deadly tumor in skin tumors and is prone to distant metastases. The incidence of melanoma has increased rapidly in the past few decades, and current trends indicate that this growth is continuing. This study was aimed to explore the molecular mechanisms of melanoma pathogenesis and discover underlying pathways and genes associated with melanoma.Methods: We used high-throughput expression data to study differential expression profiles of related genes in melanoma. The differentially expressed genes (DEGs) of melanoma in GSE15605, GSE46517, GSE7553 and the Cancer Genome Atlas (TCGA) datasets were analyzed. Differentially expressed genes (DEGs) were identified by paired t-test. Then the DEGs were performed cluster and principal component analyses and protein–protein interaction (PPI) network construction. After that, we analyzed the differential genes through bioinformatics and got hub genes. Finally, the expression of hub genes was confirmed in the TCGA databases and collected patient tissue samples.Results: Total 144 up-regulated DEGs and 16 down-regulated DEGs were identified. A total of 17 gene ontology analysis (GO) terms and 11 pathways were closely related to melanoma. Pathway of pathways in cancer was enriched in 8 DEGs, such as junction plakoglobin (JUP) and epidermal growth factor receptor (EGFR). In the PPI networks, 9 hub genes were obtained, such as loricrin (LOR), filaggrin (FLG), keratin 5 (KRT5), corneodesmosin (CDSN), desmoglein 1 (DSG1), desmoglein 3 (DSG3), keratin 1 (KRT1), involucrin (IVL) and EGFR. The pathway of pathways in cancer and its enriched DEGs may play important roles in the process of melanoma. The hub genes of DEGs may become promising melanoma candidate genes. Five key genes FLG, DSG1, DSG3, IVL and EGFR were identified in the TCGA database and melanoma tissues.Conclusions: The results suggested that FLG, DSG1, DSG3, IVL and EGFR might play important roles and potentially be valuable in the prognosis and treatment of melanoma.

Download Full-text

Identification of Differentially Expressed Genes and miRNAs Associated with Esophageal Squamous Cell Carcinoma by Integrated Analysis of Microarray Data

BioMed Research International ◽

10.1155/2020/1980921 ◽

2020 ◽

Vol 2020 ◽

pp. 1-16

Author(s):

Lemeng Zhang ◽

Jianhua Chen ◽

Tianli Cheng ◽

Hua Yang ◽

Changqie Pan ◽

...

Keyword(s):

Gene Expression ◽

Squamous Cell Carcinoma ◽

Survival Analysis ◽

Esophageal Squamous Cell Carcinoma ◽

Cell Carcinoma ◽

Differentially Expressed Genes ◽

Microarray Data ◽

Regulatory Network ◽

Differentially Expressed ◽

Pathway Enrichment

To identify candidate key genes and miRNAs associated with esophageal squamous cell carcinoma (ESCC) development and prognosis, the gene expression profiles and miRNA microarray data including GSE20347, GSE38129, GSE23400, and GSE55856 were downloaded from the Gene Expression Omnibus (GEO) database. Clinical and survival data were retrieved from The Cancer Genome Atlas (TCGA). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of differentially expressed genes (DEGs) was analyzed via DAVID, while the DEG-associated protein-protein interaction network (PPI) was constructed using the STRING database. Additionally, the miRNA target gene regulatory network and miRNA coregulatory network were constructed, using the Cytoscape software. Survival analysis and prognostic model construction were performed via the survival (version 2.42-6) and rbsurv R packages, respectively. The results showed a total of 2575, 2111, and 1205 DEGs, and 226 differentially expressed miRNAs (DEMs) were identified. Pathway enrichment analyses revealed that DEGs were mainly enriched in 36 pathways, such as the proteasome, p53, and beta-alanine metabolism pathways. Furthermore, 448 nodes and 1144 interactions were identified in the PPI network, with MYC having the highest random walk score. In addition, 7 DEMs in the microarray data, including miR-196a, miR-21, miR-205, miR-194, miR-103, miR-223, and miR-375, were found in the regulatory network. Moreover, several reported disease-related miRNAs, including miR-198a, miR-103, miR-223, miR-21, miR-194, and miR-375, were found to have common target genes with other DEMs. Survival analysis revealed that 85 DEMs were related to prognosis, among which hsa-miR-1248, hsa-miR-1291, hsa-miR-421, and hsa-miR-7-5p were used for a prognostic survival model. Taken together, this study revealed the important roles of DEGs and DEMs in ESCC development, as well as DEMs in the prognosis of ESCC. This will provide potential therapeutic targets and prognostic predictors for ESCC.

Download Full-text

PO-046 Gene expression profiling of peripheral blood mononuclear cells in young male trampoline athletes

Exercise Biochemistry Review ◽

10.14428/ebr.v1i3.10113 ◽

2018 ◽

Vol 1 (3) ◽

Author(s):

Li Gao ◽

Yong Jie Yang ◽

En Qi Li ◽

Jia Ning Mao

Keyword(s):

Gene Expression ◽

Differentially Expressed Genes ◽

Bone Health ◽

Expression Profiling ◽

Molecular Mechanisms ◽

Interaction Network ◽

Differentially Expressed ◽

Signal Pathways ◽

Healthy Controls ◽

Mineral Density

Objective Evidence indicates that physical activity influence bone health. However, the molecular mechanisms mediating the beneficial adaptations to exercise are not well understood. The purpose of this study was to examine the differentially expressed genes in PBMC between athletes and healthy controls, and to analyze the important functional genes and signal pathways that cause increased bone mineral density in athletes, in order to further reveal the molecular mechanisms of exercise promoting bone health. Methods Five professional trampoline athletes and five age-matched untrained college students participated in this study. Used the human expression Microarray V4.0 expression profiling chip to detect differentially expressed genes in the two groups, and performed KEGG Pathway analysis and application of STRING database to construct protein interaction Network; Real-Time PCR technology was used to verify the expression of some differential genes. Results Compared with healthy controls, there were significant improvement in lumbar spine bone mineral density, and 236 up-regulated as well as 265 down-regulated in serum samples of athletes. The differentially expressed genes involved 28 signal pathways, such as cell adhesion molecules. Protein interaction network showed that MYC was at the core node position. Real-time PCR results showed that the expression levels of CD40 and ITGα6 genes in the athletes were up-regulated compared with the healthy controls, the detection results were consistent with that of the gene chip. Conclusions The findings highlight that long-term high-intensity trampoline training could induce transcriptional changes in PBMC of the athletes. These data suggest that gene expression fingerprints can serve as a powerful research tool to design novel strategies for monitoring exercise. The findings of the study also provide support for the notion that PBMC could be used as a substitute to study exercise training that affects bone health.

Download Full-text

Identification of key genes and pathways associated with cholangiocarcinoma development based on weighted gene correlation network analysis

PeerJ ◽

10.7717/peerj.7968 ◽

2019 ◽

Vol 7 ◽

pp. e7968 ◽

Cited By ~ 7

Author(s):

Jingwei Liu ◽

Weixin Liu ◽

Hao Li ◽

Qiuping Deng ◽

Meiqi Yang ◽

...

Keyword(s):

Network Analysis ◽

Differentially Expressed Genes ◽

Cox Regression ◽

Nicotinamide Adenine Dinucleotide Phosphate ◽

The Cancer Genome Atlas ◽

Differentially Expressed ◽

Clinicopathological Parameters ◽

Glutathione Metabolism ◽

Kegg Analysis ◽

Key Genes

Background As the most frequently occurred tumor in biliary tract, cholangiocarcinoma (CCA) is mainly characterized by its late diagnosis and poor outcome. It is therefore urgent to identify specific genes and pathways associated with its progression and prognosis. Materials and Methods The differentially expressed genes in The Cancer Genome Atlas were analyzed to build the co-expression network by Weighted gene co-expression network analysis (WGCNA). Gene ontology (GO) as well as Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis were conducted for the selected genes. Module–clinical trait relationships were analyzed to explore the association with clinicopathological parameters. Log-rank tests and cox regression were used to identify the prognosis-related genes. Results The most related modules with CCA development were tan module containing 181 genes and salmon module with 148 genes. GO analysis suggested enrichment terms of digestion, hormone transport and secretion, epithelial cell proliferation, signal release, fibroblast activation, response to acid chemical, wnt, Nicotinamide adenine dinucleotide phosphate metabolism. KEGG analysis demonstrated 15 significantly altered pathways including glutathione metabolism, wnt, central carbon metabolism, mTOR, pancreatic secretion, protein digestion, axon guidance, retinol metabolism, insulin secretion, salivary secretion, fat digestion. Key genes of SOX2, KIT, PRSS56, WNT9A, SLC4A4, PRRG4, PANX2, PIR, RASSF8, MFSD4A, INS, RNF39, IL1R2, CST1, and PPP3CA might be potential prognostic markers for CCA, of which RNF39 and PRSS56 also showed significant correlation with clinical stage. Discussion Differentially expressed genes and key modules contributing to CCA development were identified by WGCNA. Our results offer novel insights into the characteristics in the etiology, prognosis, and treatment of CCA.

Download Full-text

Application of a Deep Matrix Factorization Model on Integrated Gene Expression Data

Current Bioinformatics ◽

10.2174/1574893614666191017094331 ◽

2020 ◽

Vol 15 (4) ◽

pp. 359-367

Author(s):

Yong-Jing Hao ◽

Mi-Xiao Hou ◽

Ying-Lian Gao ◽

Jin-Xing Liu ◽

Xiang-Zhen Kong

Keyword(s):

Gene Expression ◽

Differentially Expressed Genes ◽

Gene Expression Data ◽

Matrix Factorization ◽

Coefficient Matrix ◽

The Cancer Genome Atlas ◽

Differentially Expressed ◽

Complex Data ◽

Expression Data ◽

Factorization Model

Background: Non-negative Matrix Factorization (NMF) has been extensively used in gene expression data. However, most NMF-based methods have single-layer structures, which may achieve poor performance for complex data. Deep learning, with its carefully designed hierarchical structure, has shown significant advantages in learning data features. Objective: In bioinformatics, on the one hand, to discover differentially expressed genes in gene expression data; on the other hand, to obtain higher sample clustering results. It can provide the reference value for the prevention and treatment of cancer. Method: In this paper, we apply a deep NMF method called Deep Semi-NMF on the integrated gene expression data. In each layer, the coefficient matrix is directly decomposed into the basic and coefficient matrix of the next layer. We apply this factorization model on The Cancer Genome Atlas (TCGA) genomic data. Results: The experimental results demonstrate the superiority of Deep Semi-NMF method in identifying differentially expressed genes and clustering samples. Conclusion: The Deep Semi-NMF model decomposes a matrix into multiple matrices and multiplies them to form a matrix. It can also improve the clustering performance of samples while digging out more accurate key genes for disease treatment.

Download Full-text