Robust Significance Analysis of Microarrays by Minimum β-Divergence Method

Robust Identification of Differential Gene Expression Patterns from Multiple Transcriptomics Datasets for Early Diagnosis, Prognosis, and Therapies for Breast Cancer

Background and Objectives: Breast cancer (BC) is one of the major causes of cancer-related death in women globally. Proper identification of BC-causing hub genes (HubGs) for prognosis, diagnosis, and therapies at an earlier stage may reduce such death rates. However, most of the previous studies detected HubGs through non-robust statistical approaches that are sensitive to outlying observations. Therefore, the main objectives of this study were to explore BC-causing potential HubGs from robustness viewpoints, highlighting their early prognostic, diagnostic, and therapeutic performance. Materials and Methods: Integrated robust statistics and bioinformatics methods and databases were used to obtain the required results. Results: We robustly identified 46 common differentially expressed genes (cDEGs) between BC and control samples from three microarrays (GSE26910, GSE42568, and GSE65194) and one scRNA-seq (GSE235168) dataset. Then, we identified eight cDEGs (COL11A1, COL10A1, CD36, ACACB, CD24, PLK1, UBE2C, and PDK4) as the BC-causing HubGs by the protein-protein interaction (PPI) network analysis of cDEGs. The performance of BC and survival probability prediction models with the expressions of HubGs from two independent datasets (GSE45827 and GSE54002) and the TCGA (The Cancer Genome Atlas) database showed that our proposed HubGs might be considered as diagnostic and prognostic biomarkers, where two genes, COL11A1 and CD24, exhibit better performance. The expression analysis of HubGs by Box plots with the TCGA database in different stages of BC progression indicated their early diagnosis and prognosis ability. The HubGs set enrichment analysis with GO (Gene ontology) terms and KEGG (Kyoto Encyclopedia of Genes and Genomes) pathways disclosed some BC-causing biological processes, molecular functions, and pathways. Finally, we suggested the top-ranked six drug molecules (Suramin, Rifaximin, Telmisartan, Tukysa Tucatinib, Lynparza Olaparib, and TG.02) for the treatment of BC by molecular docking analysis with the proposed HubGs-mediated receptors. Molecular docking analysis results also showed that these drug molecules may inhibit cancer-related post-translational modification (PTM) sites (Succinylation, phosphorylation, and ubiquitination) of hub proteins. Conclusions: This study's findings might be valuable resources for diagnosis, prognosis, and therapies at an earlier stage of BC.

Robust identification of common genomic biomarkers from multiple gene expression profiles for the prognosis, diagnosis, and therapies of pancreatic cancer

Pancreatic cancer (PC) is one of the leading causes of cancer-related death globally. So, identification of potential molecular signatures is required for diagnosis, prognosis, and therapies of PC. In this study, we detected 71 common differentially expressed genes (cDEGs) between PC and control samples from four microarray gene-expression datasets (GSE15471, GSE16515, GSE71989, and GSE22780) by using robust statistical and machine learning approaches, since microarray gene-expression datasets are often contaminated by outliers due to several steps involved in the data generating processes. Then we detected 8 cDEGs (ADAM10, COL1A2, FN1, P4HB, ITGB1, ITGB5, ANXA2, and MYOF) as the PC-causing key genes (KGs) by the protein-protein interaction (PPI) network analysis. We validated the expression patterns of KGs between case and control samples by box plot analysis with the TCGA and GTEx databases. The proposed KGs showed high prognostic power with the random forest (RF) based prediction model and Kaplan-Meier-based survival probability curve. The KGs regulatory network analysis detected few transcriptional and post-transcriptional regulators for KGs. The cDEGs-set enrichment analysis revealed some crucial PC-causing molecular functions, biological processes, cellular components, and pathways that are associated with KGs. Finally, we suggested KGs-guided five repurposable drug molecules (Linsitinib, CX5461, Irinotecan, Timosaponin AIII, and Olaparib) and a new molecule (NVP-BHG712) against PC by molecular docking. The stability of the top three protein-ligand complexes was confirmed by molecular dynamic (MD) simulation studies. The cross-validation and some literature reviews also supported our findings. Therefore, the finding of this study might be useful resources to the researchers and medical doctors for diagnosis, prognosis and therapies of PC by the wet-lab validation.

Integrated Gene Expression and Methylation Analyses Identify DLL3 as a Biomarker for Prognosis of Malignant Glioma

Glioma is one of the most common neurological malignancies worldwide. Delta-like ligand 3 (DLL3), an inhibitory ligand-driven activation of the Notch pathway, has been shown to be significantly associated with overall survival in patients with glioma. Therefore, the purpose of this study was to determine whether DLL3 as a biomarker in glioma is associated with patients' clinicopathological features and prognosis. We identified differences in transcriptome and promoter methylation in the Chinese Glioma Genome Atlas (CGGA) in patients with malignant glioma with shorter (less than 1 year) and longer (greater than 3 years) survival time. Further analysis of The Cancer Genome Atlas (TCGA) revealed that four genes (DLL3, TSPAN15, RTN1, PAK7) are highly associated with patient prognosis and play an indispensable role in evolution. We chose the expression level of DLL3 in glioma patients for our study. Patients were divided into groups with low and high expression of DLL3 according to the cutoff values obtained, and Kaplan-Meier and Cox analysis were used to examine the correlation between DLL3 gene expression and patient survival. We then performed a gene set enrichment analysis (GSEA) to identify significantly enriched signaling pathways. Our results confirmed that the overall survival of patients with low DLL3 expression was significantly shorter than that of patients with high DLL3 expression. GSEA showed that the signaling pathways of the immune process and immune response, among others, were enhanced with the DLL3 low-expression phenotype. Collectively, our findings signify that DLL3 is a potent prognostic factor for glioma, which can provide a viable approach for glioma prognostic assessment and valuable insights for anti-tumor immune-targeted therapies.

Integrated Analysis of lncRNA-miRNA-mRNA ceRNA Network Identified lncRNA EPB41L4A-AS1 as a Potential Biomarker in Non-small Cell Lung Cancer

Background

Recent evidence has indicated that long non-coding RNAs (lncRNAs) can function as competing endogenous RNAs (ceRNAs) to modulate mRNAs expression by sponging microRNAs (miRNAs). However, the specific mechanism and function of lncRNA-miRNA-mRNA regulatory network in non-small cell lung cancer (NSCLC) remains unclear.

Materials and Methods

We constructed a lung cancer related lncRNA-mRNA network (LCLMN) by integrating differentially expressed genes (DEGs) with miRNA-target interactions. We further performed topological feature analysis and random walk with restart (RWR) analysis of LCLMN. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis were performed to investigate the target DEGs in LCLMN. The expression levels of significant lncRNAs in NSCLC were validated by quantitative real-time PCR (RT-qPCR). The prognostic value of the potential lncRNA was evaluated by Kaplan-Meier analysis.

Results

A total of 33 lncRNA nodes, 580 mRNA nodes and 2105 edges were identified from LCLMN. Based on functional enrichment analysis and co-expression analysis, lncRNA EPB41L4A-AS1 was demonstrated to be correlated with the tumorigenesis of NSCLC. RT-qPCR results confirmed that the expression levels of lncRNA EPB41L4A-AS1 in NSCLC tissues were downregulated compared with adjacent non-cancerous tissues. Kaplan-Meier analysis showed that high expression of lncRNA EPB41L4A-AS1 was associated with better overall survival (OS) in NSCLC patients. Further investigation identified that high expression levels of COL4A3BP, CDS2, PURA, PDCD6IP, and TMEM245 were also correlated with better OS in NSCLC patients.

Conclusion

In this study, we constructed a lncRNA-miRNA-mRNA ceRNA network to investigate potential prognostic biomarkers for NSCLC. We found that lncRNA EPB41L4A-AS1 could function as a regulator in the pathogenesis of NSCLC.

A Robust Approach for Identification of Cancer Biomarkers and Candidate Drugs

Background and objectives: Identification of cancer biomarkers that are differentially expressed (DE) between two biological conditions is an important task in many microarray studies. There exist several methods in the literature in this regards and most of these methods designed especially for unpaired samples, those are not suitable for paired samples. Furthermore, the traditional methods use p-values or fold change (FC) values to detect the DE genes. However, sometimes, p-value based results do not comply with FC based results due to the smaller pooled variance of gene expressions, which occurs when variance of each individual condition becomes smaller. There are some methods that combine both p-values and FC values to solve this problem. But, those methods also show weak performance for small sample cases in the presence of outlying expressions. To overcome this problem, in this paper, an attempt is made to propose a hybrid robust SAM-FC approach by combining rank of FC values and rank of p-values computed by SAM statistic using minimum β-divergence method, which is designed for paired samples. Materials and Methods: The proposed method introduces a weight function known as β-weight function. This weight function produces larger weights corresponding to usual and smaller weights for unusual expressions. The β-weight function plays the significant role on the performance of the proposed method. The proposed method uses β-weight function as a measure of outlier detection by setting β = 0.2. We unify both classical and robust estimates using β-weight function, such that maximum likelihood estimators (MLEs) are used in absence of outliers and minimum β-divergence estimators are used in presence of outliers to obtain reasonable p-values and FC values in the proposed method. Results: We examined the performance of proposed method in a comparison of some popular methods (t-test, SAM, LIMMA, Wilcoxon, WAD, RP, and FCROS) using both simulated and real gene expression profiles for both small and large sample cases. From the simulation and a real spike in data analysis results, we observed that the proposed method outperforms other methods for small sample cases in the presence of outliers and it keeps almost equal performance with other robust methods (Wilcoxon, RP, and FCROS) otherwise. From the head and neck cancer (HNC) gene expression dataset, the proposed method identified two additional genes (CYP3A4 and NOVA1) that are significantly enriched in linoleic acid metabolism, drug metabolism, steroid hormone biosynthesis and metabolic pathways. The survival analysis through Kaplan-Meier curve revealed that combined effect of these two genes has prognostic capability and they might be promising biomarker of HNC. Moreover, we retrieved the 12 candidate drugs based on gene interaction from glad4u and drug bank literature based gene associations. Conclusions: Using pathway analysis, disease association study, protein-protein interactions and survival analysis we found that our proposed two additional genes might be involved in the critical pathways of cancer. Furthermore, the identified drugs showed statistical significance which indicates that proteins associated with these genes might be therapeutic target in cancer.

Screening of schizophrenia associated miRNAs and the regulation of miR-320a-3p on integrin β1

Schizophrenia is one of the most severe chronic psychiatric disorders, which lacks of objective and effective diagnosis and observation indicators.In this work, the serum miRNA profiles of schizophrenic patients were analyzed. Targets of abnormal miRNAs, and their regulatory mechanisms were studied. A miRNA array was used to analyze the serum from 3 schizophrenic patients without treatment, 3 clinically cured patients and 3 healthy controls. The findings from the array were confirmed by real-time PCR in a larger cohort, including 59 patients and 60 healthy controls. The candidate miRNAs were analyzed using bioinformatics tools. Their potential targets were studied through in vitro cellular experiments.MiR-320a-3p and miR-320b were found to be down-regulated in patients compared with cured patients and controls in the miRNA array, which was also confirmed by real-time PCR in the larger cohort. Integrin β1 (ITG β1) was found to be one of the targets of miR-320a-3p. An enzyme-linked immune sorbent assay demonstrated that the ITG β1 concentration increased significantly in the patients' serum, and the in vitro study confirmed that miR-320a-3p targeted the 3' UTR of ITG β1 mRNA and reduced its expression.Our results demonstrated that the regulatory effect of miR-320a-3p on its target ITG β1 might play an important role in schizophrenia pathogenesis, which could be a potential pathway for schizophrenia diagnosis and therapy.

Robust Feature Selection Approach for Patient Classification using Gene Expression Data

Patient classification through feature selection (FS) based on gene expression data (GED) has already become popular to the research communities. T-test is the well-known statistical FS method in GED analysis. However, it produces higher false positives and lower accuracies for small sample sizes or in presence of outliers. To get rid from the shortcomings of t-test with small sample sizes, SAM has been applied in GED. But, it is highly sensitive to outliers. Recently, robust SAM using the minimum β-divergence estimators has overcome all the problems of classical t-test & SAM and it has been successfully applied for identification of differentially expressed (DE) genes. But, it was not applied in classification. Therefore, in this paper, we employ robust SAM as a feature selection approach along with classifiers for patient classification. We demonstrate the performance of the robust SAM in a comparison of classical t-test and SAM along with four popular classifiers (LDA, KNN, SVM and naive Bayes) using both simulated and real gene expression datasets. The results obtained from simulation and real data analysis confirm that the performance of the four classifiers improve with robust SAM than the classical t-test and SAM. From a real Colon cancer dataset we identified 21 additional DE genes using robust SAM that were not identified by the classical t-test or SAM. To reveal the biological functions and pathways of these 21 genes, we perform KEGG pathway enrichment analysis and found that these genes are involved in some important pathways related to cancer disease.