Transcriptomic and functional network features of lung squamous cell carcinoma through integrative analysis of GEO and TCGA data

Abstract

Lung squamous cell carcinoma (LUSC) is associated with poor clinical prognosis and lacks available targeted therapy. Novel molecules are urgently required for the diagnosis and prognosis of LUSC. Here, we conducted our data mining analysis for LUSC by integrating the differentially expressed genes acquired from Gene Expression Omnibus (GEO) database by comparing tumor tissues versus normal tissues (GSE8569, GSE21933, GSE33479, GSE33532, GSE40275, GSE62113, GSE74706) into The Cancer Genome Atlas (TCGA) database which includes 502 tumors and 49 adjacent non-tumor lung tissues. We identified intersections of 129 genes (91 up-regulated and 38 down-regulated) between GEO data and TCGA data. Based on these genes, we conducted our downstream analysis including functional enrichment analysis, protein-protein interaction, competing endogenous RNA (ceRNA) network and survival analysis. This study may provide more insight into the transcriptomic and functional features of LUSC through integrative analysis of GEO and TCGA data and suggests therapeutic targets and biomarkers for LUSC.

Figure 1

Flowchart for bioinformatics analysis of publicly available data from both GEO and TCGA databases.

Figure 2

Results from the principal component analysis for microarray studies downloaded from the GEO database. (A) Bar plots showing the proportion of variance evaluated for each of the five microarray datasets. (B) Two-dimensional plots of normal and tumor groups with the top two principal components. Horizontal and vertical axes represent the distribution of each sample within PCA1 and PCA2 respectively. PCA1: principle component 1; PCA2: principal component 2.

Figure 3

Convergence of gene expression signatures across different studies of LUSC. (A) Volcano plots showed the number of differentially expressed genes identified from each of the seven GEO datasets and after batch correction. (B) Volcano plot showed the number of differentially expressed genes in TCGA. (C) Venn diagram demonstrates the intersections of genes between GEO data and TCGA data. (D) Chromosome mapping of consensus genes.

Figure 4

The expression changes of these genes in GEO and TCGA data. (A) Heatmap of differentially expressed genes in GEO dataset coloring the samples-groups. (B) Heatmap of differentially expressed genes in TCGA dataset coloring the groups.

Figure 5

GO annotations, KEGG pathways, functional enrichment analysis and protein-protein interaction of up-regulated gene and down-regulated genes in LUSC. (A) The bubble plots showing GO and KEGG pathway enrichment data for genes that were up-regulated. (B) The bubble plots showing GO and KEGG pathway enrichment data for genes that were down-regulated. (C) Functional enrichment analysis plot. A negative z-score indicates that the activity is decreased. A positive z-score indicates that the activity is increased. (D) Protein-protein interaction network.

Figure 6

CeRNA network. LncRNA–miRNA–mRNA interactions in LUSC. Red indicates up-regulated lncRNAs, purple indicates up-regulated mRNAs, yellow stands for over-expressed miRNAs, and green means down-regulated miRNAs.

Figure 7

Survival analysis for differentially expressed genes in LUSC. Survival curves showing 20 examples of genes which were related to overall patient survival rate. P-value set for this analysis is less than 0.05.