SADLN: Self-attention based deep learning network of integrating multi-omics data for cancer subtype recognition.

Integrating multi-omics data for cancer subtype recognition is an important task in bioinformatics. Recently, deep learning has been applied to recognize the subtype of cancers. However, existing studies almost integrate the multi-omics data simply by concatenation as the single data and then learn a latent low-dimensional representation through a deep learning model, which did not consider the distribution differently of omics data. Moreover, these methods ignore the relationship of samples. To tackle these problems, we proposed SADLN: A self-attention based deep learning network of integrating multi-omics data for cancer subtype recognition. SADLN combined encoder, self-attention, decoder, and discriminator into a unified framework, which can not only integrate multi-omics data but also adaptively model the sample's relationship for learning an accurately latent low-dimensional representation. With the integrated representation learned from the network, SADLN used Gaussian Mixture Model to identify cancer subtypes. Experiments on ten cancer datasets of TCGA demonstrated the advantages of SADLN compared to ten methods. The Self-Attention Based Deep Learning Network (SADLN) is an effective method of integrating multi-omics data for cancer subtype recognition.

Validating HMMR Expression and Its Prognostic Significance in Lung Adenocarcinoma Based on Data Mining and Bioinformatics Methods.

Hyaluronic acid-mediated motility receptor (HMMR), a tumor-related gene, plays a vital role in the occurrence and progression of various cancers. This research is aimed to reveal the effect of HMMR in lung adenocarcinoma (LUAD). We first obtained the gene expression profiles and clinical data of patients with LUAD from The Cancer Genome Atlas (TCGA) database. Then, based on the TCGA cohort, the HMMR expression difference between LUAD tissues and nontumor tissues was detected and verified with public tissue microarrays (TMAs), clinical LUAD specimen cohort, and Gene Expression Omnibus (GEO) cohort. Logistic regression analysis and chi-square test were adopted to study the correlation between HMMR expression and clinicopathological parameters. The effect of HMMR expression on survival was evaluated by Kaplan-Meier survival analysis and using the Cox regression model. Furthermore, Gene Set Enrichment Analysis (GSEA) was utilized to screen out signaling pathways related to LUAD and the co-expression analysis was employed to build the protein-protein interaction (PPI) network. The HMMR expression level in LUAD tissues was dramatically higher than that in nontumor tissues. Logistic regression analysis and chi-square test demonstrated that the high HMMR expression in LUAD has relation with gender, pathological stage, T classification, lymph node metastasis, and distant metastasis. The Kaplan-Meier curve suggested a poor prognosis for LUAD patients with high HMMR expression. Multivariate analysis implied that the high HMMR expression was a vital independent predictor of poor overall survival (OS). GSEA indicated that a total of 15 signaling pathways were enriched in samples with the high HMMR expression phenotype. The PPI network gave 10 genes co-expressed with HMMR. HMMR may be an oncogene in LUAD and is expected to become a potential prognostic indicator and therapeutic target for LUAD.

A Novel circRNA-miRNA-mRNA Hub Regulatory Network in Lung Adenocarcinoma.

The growing evidence suggests that circular RNAs (circRNAs) have significant associations with tumor occurrence and progression, yet the regulatory mechanism of circRNAs in lung adenocarcinoma (LUAD) remains unclear. This study clarified the potentially regulatory network and functional mechanism of circRNAs in LUAD. The expression data of circRNAs, microRNAs (miRNAs), and messenger RNAs (mRNAs) were obtained from the Gene Expression Omnibus (GEO) database. Relying on GSE101586, GSE101684, and GSE112214, we identified differentially expressed circRNAs (DEcircRNAs). Depending on GSE135918 and GSE32863, we screened out differentially expressed miRNAs (DEmiRNAs) and mRNAs (DEmRNAs), respectively. Then, a novel competing endogenous RNA (ceRNA) regulatory network related to LUAD was constructed. We also revealed biological processes and signal pathways regulated by these DEcircRNAs. Based on gene expression data and survival information of LUAD patients in The Cancer Genome Atlas (TCGA) and GEO, we implemented survival analysis to select DEmRNAs related to prognosis and build a novel circRNA-miRNA-mRNA hub regulatory network. Meanwhile, quantitative real-time PCR (qRT-PCR) was utilized to validate DEcircRNAs in the ceRNA hub regulatory network. As a result, a total of 8 DEcircRNAs, 19 DEmiRNAs, and 85 DEmRNAs were identified. The novel ceRNA regulatory network included 5 circRNAs, 8 miRNAs, and 22 mRNAs. The final ceRNA hub regulatory network contained two circRNAs, two miRNAs, and two mRNAs. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses indicated that the five DEcircRNAs may affect LUAD onset and progression through Wnt signaling pathway and Hippo signaling pathway. All in all, this study revealed the regulatory network and functional mechanism of circRNA-related ceRNAs in LUAD.

Differential Expression and Bioinformatics Analysis of circRNA in Non-small Cell Lung Cancer.

Circular RNA (CircRNA) plays an important role in tumorigenesis and progression of non-small cell lung cancer (NSCLC), but the pathogenesis of NSCLC caused by circRNA has not been fully elucidated. This study aimed to investigate differentially expressed circRNAs and identify the underlying pathogenesis hub genes of NSCLC by comprehensive bioinformatics analysis. Data of gene expression microarrays (GSE101586, GSE101684, and GSE112214) were downloaded from Gene Expression Omnibus (GEO) database. Differentially expressed circRNAs (DECs) were obtained by the "limma" package of R programs and the overlapping operation was implemented of DECs. CircBase database and Cancer-Specific CircRNA database (CSCD) were used to find miRNAs binding to DECs. Target genes of the found miRNAs were identified utilizing Perl programs based on miRDB, miRTarBase, and TargetScan databases. Functional and enrichment analyses of selected target genes were performing using the "cluster profiler" package. Protein-protein interaction (PPI) network was constructed by the Search Tool for the STRING database and module analysis of selected hub genes was performed by Cytoscape 3.7.1. Survival analysis of hub genes were performed by Gene Expression Profiling Interactive Analysis (GEPIA). Respectively, 1 DEC, 249 DECs, and 101 DECs were identified in GSE101586, GSE101684, and GSE112214. A total of eight overlapped circRNAs, 43 miRNAs and 427 target genes were identified. Gene Ontology (GO) enrichment analysis showed these target genes were enriched in biological processes of regulation of histone methylation, Ras protein signal transduction and covalent chromatin modification etc. Pathway enrichment analysis showed these target genes are mainly involved in AMPK signaling pathway, signaling pathways regulating pluripotency of stem cells and insulin signaling pathway etc. A PPI network was constructed based on 427 target genes of the 43 miRNAs. Ten hub genes were found, of which the expression of MYLIP, GAN, and CDC27 were significantly related to NSCLC patient prognosis. Our study provide a deeper understanding the circRNAs-miRNAs-target genes by bioinformatics analysis, which may provide novel insights for unraveling pathogenesis of NSCLC. MYLIP, GAN, and CDC27 genes might serve as novel biomarker for precise treatment and prognosis of NSCLC in the future.