A novel graph-based k-partitioning approach improves the detection of gene-gene correlations by single-cell RNA sequencing

Background Gene expression is regulated by transcription factors, cofactors, and epigenetic mechanisms. Coexpressed genes indicate similar functional categories and gene networks. Detecting gene-gene coexpression is important for understanding the underlying mechanisms of cellular function and human diseases. A common practice of identifying coexpressed genes is to test the correlation of expression in a set of genes. In single-cell RNA-seq data, an important challenge is the abundance of zero values, so-called “dropout”, which results in biased estimation of gene-gene correlations for downstream analyses. In recent years, efforts have been made to recover coexpressed genes in scRNA-seq data. Here, our goal is to detect coexpressed gene pairs to reduce the “dropout” effect in scRNA-seq data using a novel graph-based k-partitioning method by merging transcriptomically similar cells. Results We observed that the number of zero values was reduced among the merged transcriptomically similar cell clusters. Motivated by this observation, we leveraged a graph-based algorithm and develop an R package, scCorr, to recover the missing gene-gene correlation in scRNA-seq data that enables the reliable acquisition of cluster-based gene-gene correlations in three independent scRNA-seq datasets. The graphically partitioned cell clusters did not change the local cell community. For example, in scRNA-seq data from peripheral blood mononuclear cells (PBMCs), the gene-gene correlation estimated by scCorr outperformed the correlation estimated by the nonclustering method. Among 85 correlated gene pairs in a set of 100 clusters, scCorr detected 71 gene pairs, while the nonclustering method detected only 4 pairs of a dataset from PBMCs. The performance of scCorr was comparable to those of three previously published methods. As an example of downstream analysis using scCorr, we show that scCorr accurately identified a known cell type (i.e., CD4+ T cells) in PBMCs with a receiver operating characteristic area under the curve of 0.96. Conclusions Our results demonstrate that scCorr is a robust and reliable graph-based method for identifying correlated gene pairs, which is fundamental to network construction, gene-gene interaction, and cellular omic analyses. scCorr can be quickly and easily implemented to minimize zero values in scRNA-seq analysis and is freely available at https://github.com/CBIIT-CGBB/scCorr.

Transcriptomic and physiological analysis identifies a gene network module highly associated with brassinosteroid regulation in hybrid sweetgum tissues differing in the capability of somatic embryogenesis

Somatic embryogenesis is a preferred method for large-scale production of forest trees due to its high propagation efficiency. In this study, hybrid sweetgum leaves with phase changes from mature to embryogenic state were selected as experimental material to study somatic embryo initiation. Embryogenicity ranged from high to low, i.e. from 45%, 25%, and 12.5% to 0, with the samples of embryogenic callus (EC), whiten leaf edge (WLI), whiten leaf (WLII), and green leaf (GL) respectively. High correlations existed between embryogenicity and endogenous brassinosteroids (BRs) (r = 0.95, p < 0.05). Similarly, concentrations of endogenous BRs of the sample set correlated positively (r = 0.93, 0.99, 0.87, 0.99, 0.96 respectively, P < 0.05) to expression of somatic embryo (SE)-related genes, i.e. BBM, LEC2, ABI3, PLT2, and WOX2. Hierarchical cluster and weighted gene coexpression network analysis identified modules of coexpressed genes and network in 4820 differentially expressed genes (DEGs) from All-BR-Regulated Genes (ABRG). Moreover, exogenously-supplemented epiBR, together with 2,4-D and 6-BA, increased embryogenicity of GL-sourced callus, and expression of SE- and auxin-related genes, while brassinazole (BRZ), a BR biosynthesis inhibitor, reduced embryogenicity. Evidences obtained in this study revealed that BRs involved in phase change of leaf explants and may function in regulating gene expression and enhancing auxin effects. This study successfully established protocols for inducing somatic embryogenesis from leaf explants in hybrid sweetgum, which could facilitate the propagation process greatly, and provide theoretical basis for manipulating SE competence of explants in ornamental woody plants.

Comprehensive Analysis of Myoferlin in Human Pancreatic Cancer via Bioinformatics

Pancreatic cancer is the fourth leading cause of cancer-related death and urgently needs biomarkers for clinical diagnosis and prognosis. It has been reported that myoferlin (MYOF) is implicated in the regulation of proliferation, invasion, and migration of tumor cells in many cancers including pancreatic cancer. To confirm the prognostic value of MYOF in pancreatic cancer, a comprehensive cancer versus healthy people analysis was conducted using public data. MYOF mRNA expression levels were compared in many kinds of cancers including pancreatic cancer via the Oncomine and Gene Expression Profiling Interactive Analysis (GEPIA) databases. The results have shown that MYOF mRNA expression levels were upregulated in most types of cancers, especially in pancreatic cancer, compared with healthy people’s tissues. Data from the Cancer Cell Line Encyclopedia (CCLE) and European Bioinformatics Institute (EMBL-EML) database also revealed that MYOF mRNA is highly expressed in most cancer cells, particularly in pancreatic cancer cell lines. Furthermore, the prognostic value of MYOF was evaluated using GEPIA and Long-term Outcome and Gene Expression Profiling Database of pan-cancers (LOGpc) database. Higher expression of MYOF was associated with poorer overall survival, especially in the lower stage and lower grade. Coexpressed genes, possible regulators, and the correlation between MYOF expressions were analyzed via the GEPIA and LinkedOmics database. Nineteen coexpressed genes were identified, and most of these genes were related to cancer. The Tumor Immune Estimation Resource (TIMER) database was used to analyze the correlation between MYOF and immune response. Notably, we found that MYOF might have a potential novel immune regulatory role in tumor immunity. These results support that MYOF is a candidate prognostic biomarker for pancreatic cancer, which calls for further genomics research of pancreatic cancer and deeply functional studies on MYOF.

Genetic insights into the regulatory pathways for continuous flowering in a unique orchid Arundina graminifolia

Background Manipulation of flowering time and frequency of blooming is key to enhancing the ornamental value of orchids. Arundina graminifolia is a unique orchid that flowers year round, although the molecular basis of this flowering pattern remains poorly understood. Results We compared the A. graminifolia transcriptome across tissue types and floral developmental stages to elucidate important genetic regulators of flowering and hormones. Clustering analyses identified modules specific to floral transition and floral morphogenesis, providing a set of candidate regulators for the floral initiation and timing. Among candidate floral homeotic genes, the expression of two FT genes was positively correlated with flower development. Assessment of the endogenous hormone levels and qRT-PCR analysis of 32 pathway-responsive genes supported a role for the regulatory networks in floral bud control in A. graminifolia. Moreover, WGCNA showed that flowering control can be delineated by modules of coexpressed genes; especially, MEgreen presented group of genes specific to flowering. Conclusions Candidate gene selection coupled with hormonal regulators brings a robust source to understand the intricate molecular regulation of flowering in precious orchids.

Systematic analyses of the role of prognostic and immunological EIF3A, a reader protein, in clear cell renal cell carcinoma

Background Eukaryotic initiation factor 3a (EIF3A), a “reader” protein for RNA methylation, has been found to be involved in promoting tumorigenesis in a variety of cancers. The impact of EIF3A in clear cell renal cell carcinoma (ccRCC) has yet to be reported. This study aimed to identify the prognostic value of EIF3A in ccRCC and investigate the relationship between EIF3A expression and immune infiltration. Methods We collected 29 m6A-related mRNA data and clinicopathological parameters from The Cancer Genome Atlas (TCGA) database. Logistic regression analyses were used to analyse the correlation between EIF3A expression and clinical characteristics. Immunohistochemistry (IHC) was applied to examine EIF3A levels in normal and ccRCC tissues. Univariate and multivariate analyses were conducted to recognize independent factors associated with overall survival (OS) and disease-free survival (DFS). The nomogram aimed to predict the 1-, 3- and 5-year survival probabilities. Gene set enrichment analysis (GSEA) was carried out to determine the potential functions and related signalling pathways of EIF3A expression. To investigate EIF3A of coexpressed genes, we used LinkedOmics, and the results were subjected to enrichment analysis. Simultaneously, LinkedOmics and STRING datasets were used to identify EIF3A coexpressed genes that were visualized via Cytoscape. Finally, we evaluated whether EIF3A expression correlated with the infiltration of immune cells and the expression of marker genes in ccRCC by Tumour Immune Estimation Resource (TIMER) and Gene Expression Profiling Interactive Analysis (GEPIA). Result EIF3A expression was significantly different between ccRCC tissues and normal tissues. EIF3A expression was correlated with poor prognostic clinicopathological factors, and K–M analyses revealed that low EIF3A expression was correlated with a poor prognosis. The results of univariate and multivariate analyses proved that EIF3A was a prognostic factor in ccRCC patients. GSEA results indicated that EIF3A high expression was enriched in the renal cell carcinoma pathway. EIF3A expression was significantly positively correlated with B cells, CD8 + T cells, CD4 + T cells, neutrophils, macrophages, and dendritic cells. Furthermore, EIF3A expression was associated with most marker genes of immune cells. Conclusions EIF3A could serve as a potential biomarker for prognostic and diagnostic stratification of ccRCC and is related to immune cell infiltrates.

The Alterations and Potential Roles of MCMs in Breast Cancer

The minichromosome maintenance (MCM) protein family plays a key role in eukaryotic DNA replication and has been confirmed to be associated with the occurrence and progression of many tumors. However, the expression levels, functions, and prognostic values of MCMs in breast cancer (BC) have not been clearly and systematically explained. In this article, we studied the transcriptional levels of MCMs in BC based on the Oncomine database. Kaplan-Meier plotter was used to analyze prognostic value of MCMs in human BC patients. Furthermore, we constructed a MCM coexpression gene network and performed functional annotation analysis through DAVID to reveal the functions of MCMs and coexpressed genes. The data showed that the expression of MCM2–8 and MCM10 but not MCM1 and MCM9 was upregulated in BC. Kaplan-Meier plotter analysis revealed that high transcriptional levels of MCM2, MCM4–7, and MCM10 were significantly related to low relapse-free survival (RFS) in BC patients. In contrast, high levels of MCM1 and MCM9 predicted high RFS for BC patients. This study suggests that MCM2, MCM4–7, and MCM10 possess great potential to be valuable prognostic biomarkers for BC and that MCM1 and MCM9 may serve as potential treatment targets for BC patients.

Bioinformatics analysis of the transcriptional expression of minichromosome maintenance proteins as potential indicators of survival in patients with cervical cancer

Background As major regulators of DNA replication in eukaryotes, minichromosome maintenance (MCM) proteins play an important role in the initiation and extension of DNA replication. MCMs and their related genes may be new markers of cell proliferation activity, which is of great significance for the diagnosis and prognosis of cervical cancer. Methods To explore the role of MCMs and their related genes in cervical cancer, various bioinformatics methods were performed. First, the ONCOMINE and UALCAN databases were used to analyze the mRNA expression of different MCMs. The Human Protein Atlas database was used to analyze the protein expression of MCMs in normal and tumor tissues. The potential clinical value of MCMs was evaluated using the UALCAN, Kaplan-Meier plotter and cBioPortal databases. Then, the related genes and key coexpressed genes of MCMs were screened using GEPIA2 and cBioPortal analysis. For these genes, we used Metascape and the DAVID database to perform Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses, construct the related molecular interaction network, and obtain the key subnetworks and related hub genes. The Kaplan-Meier plotter database was used for survival analysis of cervical cancer patients to evaluate and predict the potential clinical value of the hub genes. Moreover, multiple gene comparisons of the expression of MCMs and related genes in different cancer types also showed the clinical significance of these potential targets. Results The mRNA and protein expression of MCMs increased in tumor tissue. Overexpression of MCM2/3/4/5/6/7/8/10 was found to be significantly associated with clinical cancer stage. Higher mRNA expression levels of MCM3/5/6/7/8 were found to be significantly associated with longer overall survival, and higher mRNA expression of MCM2/3/4/5/6/7/8 was associated with favorable OS. In addition, a high mutation rate of MCMs (71%) was observed. MCM2, MCM4, MCM8, MCM3 and MCM7 were the five genes with the most genetic alterations. In addition, the coexpressed genes and related genes of MCMs were successfully screened for enrichment analysis. These genes were significantly enriched in important pathways, such as the DNA replication, cell cycle, mismatch repair, spliceosome, and Fanconi anemia pathways. A protein-protein interaction network was successfully constructed, and a total of 13 hub genes (CDC45, ORC1, RPA1, CDT1, TARDBP, RBMX, SRSF3, SRSF1, RFC5, RFC2, MSH6, DTL, and MSH2) from 4 key subnetworks were obtained. These genes and MCM2/3/4/5/6/7/8 might have potential clinical value for the survival and prognosis of cervical cancer patients. Conclusions These findings promoted the understanding of the MCM protein family and clinically related molecular targets for cervical epithelial neoplasia and cervical cancer. Our results were helpful to evaluate the potential clinical value of MCMs and related genes in patients with cervical cancer.

Tumor Purity Coexpressed Genes Related to Immune Microenvironment and Clinical Outcomes of Lung Adenocarcinoma

Purpose. Lung cancer tissue includes tumor tissue, stromal cells, immune cells, and epithelial cells. These nontumor cells dilute the tumor purity in lung cancer tissues. Tumor purity plays an essential role in the immune response to lung cancer. At present, the biological processes related to the purity of lung cancer tumors remains unclear. Methods. We measured tumor purity in 486 lung carcinoma tissues from TCGA-LUAD FPKM by using the “estimate” R package. Lung carcinoma tumor mutation burden was calculated by analyzing TCGA single nucleotide polymorphism data. The immune cell proportion was also evacuated via the CIBERSORT method. Lung carcinoma samples with P < 0.05 were considered significant. Based on the tumor purity and lung carcinoma gene matrix, we performed weighted gene coexpression network analysis (WGCNA), and the tumor purity-related module was identified. Then, we analyzed the functions of the factors involved in the module. We screened the coexpressed factors related to clinical outcome and immunophenotype. Finally, expression levels of these factors were measured at tissue and single-cell levels. Results. A lung cancer tumor purity correlated coexpression network was determined. Five coexpressed genes (CD4, CD53, EVI2B, PLEK, and SASH3) were identified as tumor purity coexpressed genes that negatively correlated with tumor purity. Because the factors in the coexpression network often participate in similar biological processes, we found that CD4, CD53, EVI2B, PLEK, and SASH3 were most related to positive regulation of cytokine production and interleukin−2 production through functional enrichment. In a clinical phenotype analysis, we found that these five factors can be used as independent prognostic risk factors. We found that these factors were significantly negatively correlated with tumor purity and positively correlated with the immune score in the immunophenotyping analysis. Using GSEA analysis, we found that the antigen processing and presentation pathway were related to the five tumor coexpressed genes mentioned above. SASH3 and CD53 were used to conduct a prognostic model based on the interaction analysis of the Support Vector Machine and the Least Absolute Shrinkage and Selection Operator. SASH3 was verified to be related to CD8A using a single-cell analysis. Conclusion. Tumor purity-related coexpression factors in the tumor microenvironment have essential clinical, genomic, and biological significance in lung cancer. These coexpression factors (SASH3 and CD53) can be used to classify tumor purity phenotypes and to predict clinical outcomes.

Identification of CD8+ T Cell-Related Genes: Correlations with Immune Phenotypes and Outcomes of Liver Cancer

Purpose. Treatment outcomes for advanced liver cancer are poor. Immunotherapy is a treatment strategy that has been widely used to treat other cancers. Studies have shown that CD8+ T lymphocytes are essential factors affecting the efficacy of immunotherapy. We used computational biology methods to determine the coexpressed gene network that promotes CD8+ T lymphocyte infiltration. Method. We obtained the liver cancer gene matrix and clinical follow-up information data from TCGA liver hepatocellular carcinoma FPKM. We obtained single nucleotide polymorphism (SNP) data to evaluate the tumor mutation burden. The “estimate” package and the CIBERSORT algorithm were used to evaluate tumor purity and the proportion of CD8+ T lymphocytes in the liver cancer cohort. We used the gene expression matrix of liver cancer and the relative proportion of CD8+ T lymphocytes as input files and performed WGCNA based on this analysis. The weighted coexpression network identified the most CD8+ T lymphocyte-related coexpression modules in liver cancer. Then, we analyzed the biological processes involved in the module. We determined the coexpression module with CD8+ T lymphocyte infiltration in terms of data and function. We then screened the factors in the coexpression module correlated with CD8+ T lymphocyte content greater than 0.4. Finally, the expression levels of these factors were verified at the protein level using immunohistochemistry and single-cell sequencing. Results. We determined the CD8+ T lymphocyte proportions that correlated with coexpression networks. Four coexpressed genes (C1QC, CD3D, GZMA, and PSMB9) were identified as CD8+ T cell coexpression genes that promoted infiltration of CD8+ T cells. Because the factors in the coexpression network often participate in similar biological processes, we found that these factors were most related to antigen processing and presentation of peptide antigen through functional enrichment. In the clinical phenotype analysis, we found that 18 factors can be used as independent prognostic protective factors. We found that these factors were significantly negatively correlated with tumor purity and negatively correlated with M2 macrophages in the immunophenotyping analysis. Using immunohistochemistry and single-cell sequencing analysis, we found that CD3D antibody staining was weaker in tumor tissues than normal tissues and was related to CD8+ T cells. Conclusion. These coexpressed genes were positively related to the high infiltration proportion of CD8+ T lymphocytes in an antigen presentation process. The biological process might provide new directions for patients who are insensitive to immune therapy.

PD-1 Coexpression Gene Analysis and the Regulatory Network in Endometrial Cancer Based on Bioinformatics Analysis

Gynecological malignancies are tumors of the female reproductive system, mainly cervical cancer, endometrial cancer, and ovarian cancer. Endometrial cancer (EC) is the most common gynecological malignant tumor in developed countries. The aim of this study was to construct a network of programmed cell death protein 1 (PD-1) coexpressed genes through bioinformatics analysis and screen the potential biomarkers of PD-1 in endometrial cancer. In addition, genes and pathways involved in PD-1 and modulating tumor immune status were identified. We select the EC transcriptomic dataset in TCGA to retrieve gene sets on the cBioPortal platform, and the PD-1 coexpressed genes were obtained on the platform. GO and KEGG enrichment analysis of coexpressed genes was performed using the DAVID database. The target protein-protein interaction (PPI) network was constructed using Cytoscape 3.7.1 software, and the hub genes were then screened. A total of 976 coexpression genes were obtained. The enrichment analysis showed that PD-1 coexpressed genes were significantly enriched in overall components of the cell structure, the interaction of cytokines with cytokine receptors, chemokine signaling pathways, and cell adhesion molecules (CAMs). Ten hub genes were obtained by node degree analysis. CD3E gene is involved in the prognosis and immune process of EC, and the expression level is related to PD-1 (Pearson correlation coefficient is 0.82, P < 0.01 ). Patients with low CD3E gene expression in EC have a poor prognosis. The coexpression hub genes of PD-1 are related to immunity, in which CD3E is a prognostic marker that is involved in the PD-1/PD-L1-induced tumor immune escape. This study provides a new area to study the mechanism of PD-1/PD-L1 in EC and the precise treatment with targeted drugs.