Identification of key genes in colorectal cancer diagnosis by co-expression analysis weighted gene co-expression network analysis

Reinventing gene expression connectivity through regulatory and spatial structural empowerment via principal node aggregation graph neural network

The intricacies of the human genome, manifested as a complex network of genes, transcend conventional representations in text or numerical matrices. The intricate gene-to-gene relationships inherent in this complexity find a more suitable depiction in graph structures. In the pursuit of predicting gene expression, an endeavor shared by predecessors like the L1000 and Enformer methods, we introduce a novel spatial graph-neural network (GNN) approach. This innovative strategy incorporates graph features, encompassing both regulatory and structural elements. The regulatory elements include pair-wise gene correlation, biological pathways, protein-protein interaction networks, and transcription factor regulation. The spatial structural elements include chromosomal distance, histone modification and Hi-C inferred 3D genomic features. Principal Node Aggregation models, validated independently, emerge as frontrunners, demonstrating superior performance compared to traditional regression and other deep learning models. By embracing the spatial GNN paradigm, our method significantly advances the description of the intricate network of gene interactions, surpassing the performance, predictable scope, and initial requirements set by previous methods.

Prediction of novel biomarkers for gastric intestinal metaplasia and gastric adenocarcinoma using bioinformatics analysis

Background & aim

The histologic and molecular changes from intestinal metaplasia (IM) to gastric cancer (GC) have not been fully characterized. The present study sought to identify potential alterations in signaling pathways in IM and GC to predict disease progression; these alterations can be considered therapeutic targets.

Materials & methods

Seven gene expression profiles were selected from the GEO database. Discriminate differentially expressed genes (DEGs) were analyzed by EnrichR. The STRING database, Cytoscape, Gene Expression Profiling Interactive Analysis (GEPIA), cBioPortal, NetworkAnalyst, MirWalk database, OncomiR, and bipartite miRNA‒mRNA correlation network was used for downstream analyses of selected module genes.

Results

Analyses revealed that extracellular matrix-receptor interactions (ITGB1, COL1A1, COL1A2, COL4A1, FN1, COL6A3, and THBS2) in GC and PPAR signaling pathway interactions (FABP1, APOC3, APOA1, HMGCS2, and PPARA and PCK1) in IM may play key roles in both the carcinogenesis and progression of underlying GC from intestinal metaplasia. IM enrichment indicated that this is closely related to digestion and absorption. The TF-hub gene regulatory network revealed that AR, TCF4, SALL4, and ESR1 were more important for hub gene expression. It was revealed that the development and prediction of GC may be affected by hsa-miR-29. It was found that PTGR1, C1orf115, CRYL1, ALDOB, and SULT1B1 were downregulated in GC and upregulated in IM. Therefore, they might have tumor suppressor activity in GC progression.

Conclusion

New potential biomarkers and pathways involved in GC and IM were identified that are important for the transformation of GC from IM to adenocarcinoma and can be therapeutic targets for GC.

SG-Transunet: A segmentation-guided Transformer U-Net model for KRAS gene mutation status identification in colorectal cancer

Accurately identifying the Kirsten rat sarcoma virus (KRAS) gene mutation status in colorectal cancer (CRC) patients can assist doctors in deciding whether to use specific targeted drugs for treatment. Although deep learning methods are popular, they are often affected by redundant features from non-lesion areas. Moreover, existing methods commonly extract spatial features from imaging data, which neglect important frequency domain features and may degrade the performance of KRAS gene mutation status identification. To address this deficiency, we propose a segmentation-guided Transformer U-Net (SG-Transunet) model for KRAS gene mutation status identification in CRC. Integrating the strength of convolutional neural networks (CNNs) and Transformers, SG-Transunet offers a unique approach for both lesion segmentation and KRAS mutation status identification. Specifically, for precise lesion localization, we employ an encoder-decoder to obtain segmentation results and guide the KRAS gene mutation status identification task. Subsequently, a frequency domain supplement block is designed to capture frequency domain features, integrating it with high-level spatial features extracted in the encoding path to derive advanced spatial-frequency domain features. Furthermore, we introduce a pre-trained Xception block to mitigate the risk of overfitting associated with small-scale datasets. Following this, an aggregate attention module is devised to consolidate spatial-frequency domain features with global information extracted by the Transformer at shallow and deep levels, thereby enhancing feature discriminability. Finally, we propose a mutual-constrained loss function that simultaneously constrains the segmentation mask acquisition and gene status identification process. Experimental results demonstrate the superior performance of SG-Transunet over state-of-the-art methods in discriminating KRAS gene mutation status.

Potential biomarkers: The hypomethylation of cg18949415 and cg22193385 sites in colon adenocarcinoma

Overall cancer hypomethylation had been identified in the past, but it is not clear exactly which hypomethylation site is the more important for the occurrence of cancer. To identify key hypomethylation sites, we studied the effect of hypomethylation in twelve regions on gene expression in colon adenocarcinoma (COAD). The key DNA methylation sites of cg18949415, cg22193385 and important genes of C6orf223, KRT7 were found by constructing a prognostic model, survival analysis and random combination prediction a series of in-depth systematic calculations and analyses, and the results were validated by GEO database, immune microenvironment, drug and functional enrichment analysis. Based on the expression values of C6orf223, KRT7 genes and the DNA methylation values of cg18949415, cg22193385 sites, the least diversity increment algorithm were used to predict COAD and normal sample. The 100 % reliability and 97.12 % correctness of predicting tumor samples were obtained in jackknife test. Moreover, we found that C6orf223 gene, cg18949415 site play a more important role than KRT7 gene, cg22193385 site in COAD. In addition, we investigate the impact of key methylation sites on three-dimensional chromatin structure. Our results will be help for experimental studies and may be an epigenetic biomarker for COAD.

Analysis of the potential pyroptosis mechanism in psoriasis and experimental validation of NLRP3 in vitro and in vivo

Pyroptosis provides new perspectives on the mechanisms underlying psoriasis and the development of new treatment strategies. Here, we aimed to identify pyroptosis-related genes (PRGs) involved in the pathogenesis and progression of psoriasis. Based on the inclusion/exclusion criteria, three gene datasets were downloaded from the Gene Expression Omnibus (GEO) database. Differential gene expression, weighted gene co-expression network analysis (WGCNA), and functional enrichment analyses were performed to identify candidate PRGs for psoriasis. Least absolute shrinkage and selection operator (LASSO) regression was used to identify hub genes, and receiver operating characteristic (ROC) curves were used to determine the clinical value of the hub genes. Imiquimod-inducedpsoriasis-like mice and lipopolysaccharide (LPS)-induced RAW 264.7 cells were employed to verify the pro-inflammatory factors that may drive changes in pyroptosis. In total, 159 skin samples were analysed, and a total of 21 common targets were obtained by crossing PRGs with all the differentially expressed genes (DEGs) in different disease states. 11 genes were identified via LASSO screening. Similarly, the last six PRGs biomarkers and the green module genes were screened. All hub genes with an area under the ROC curve > 0.5 were intersected, and NLRP3 was identified. NLRP3 expression was elevated in imiquimod-induced psoriatic lesions in mice and LPS-stimulated RAW 264.7 cells. The mice exhibited reduced psoriasis area and severity index scores, hyperproliferation, and inflammation after treatment with MCC950 (a specific inhibitor of NLRP3). MCC950 decreased IL-1β, IL-6, and TNF-α mRNA expression, and NLRP3 and p-p65 protein levels in LPS-stimulated RAW 264.7 cells. Our study indicates that NLRP3 may be a promising therapeutic target for the treatment of psoriasis.

Potential Signature Therapeutic Biomarkers TOP2A, MAD2L1, and CDK1 in Colorectal Cancer: A Systems Biomedicine-Based Approach

Colorectal cancer is the third deadliest and fourth most diagnosed cancer. It is heterogeneously driven by varied mutations and mutagens, and thus, it is challenging for targeted therapy. The rapid advancement of high-throughput technology presents considerable opportunities for discovering new colon cancer biomarkers. In the present study, we have explored and identified the biomarkers based on molecular interactions. We curated cancer datasets that were not micro-dissected and performed gene expression analysis. The protein-protein interactions were curated, and a network was constructed for the up-regulated genes. The hub genes were analyzed using 12 different topological parameters. The correlation analysis selected TOP2A, CDK1, CCNB1, AURKA, and MAD2L1 as hub genes. Further, survival analysis was performed to determine the effectiveness of the hub gene on the patient's survival rate. Our findings explore various transcription factors such as E2F4, FOXM1, E2F6, MAX, and SIN3A, along with kinases CSNK2A1, MAPK14, CDK1, CDK4, and CDK2, as potential molecular signatures and aid researchers in understanding the pathophysiological mechanisms underlying CRC development and thus providing novel therapeutic and diagnostic recourse. Furthermore, investigating miRNAs, we focused on hsa-miR-215-5p, hsa-miR-192-5p, and hsa-miR-193b-3p due to their observed impact on a diverse set of colorectal cancer genes. Thereby, the current approach brings into light CRC- related genes at the RNA and protein levels that can potentially act as novel biomarkers opening doors to diagnostic and treatment purposes.

Identification of GUCA2A and COL3A1 as prognostic biomarkers in colorectal cancer by integrating analysis of RNA-Seq data and qRT-PCR validation

By 2030, it is anticipated that there will be 2.2 million new instances of colorectal cancer worldwide, along with 1.1 million yearly deaths. Therefore, it is critical to develop novel biomarkers that could help in CRC early detection. We performed an integrated analysis of four RNA-Seq data sets and TCGA datasets in this study to find novel biomarkers for diagnostic, prediction, and as potential therapeutic for this malignancy, as well as to determine the molecular mechanisms of CRC carcinogenesis. Four RNA-Seq datasets of colorectal cancer were downloaded from the Sequence Read Archive (SRA) database. The metaSeq package was used to integrate differentially expressed genes (DEGs). The protein-protein interaction (PPI) network of the DEGs was constructed using the string platform, and hub genes were identified using the cytoscape software. The gene ontology and KEGG pathway enrichment analysis were performed using enrichR package. Gene diagnostic sensitivity and its association to clinicopathological characteristics were demonstrated by statistical approaches. By using qRT-PCR, GUCA2A and COL3A1 were examined in colon cancer and rectal cancer. We identified 5037 differentially expressed genes, including (4752 upregulated, 285 downregulated) across the studies between CRC and normal tissues. Gene ontology and KEGG pathway analyses showed that the highest proportion of up-regulated DEGs was involved in RNA binding and RNA transport. Integral component of plasma membrane and mineral absorption pathways were identified as containing down-regulated DEGs. Similar expression patterns for GUCA2A and COL3A1 were seen in qRT-PCR and integrated RNA-Seq analysis. Additionally, this study demonstrated that GUCA2A and COL3A1 may play a significant role in the development of CRC.

A macrophage related signature for predicting prognosis and drug sensitivity in ovarian cancer based on integrative machine learning

BACKGROUND

Ovarian cancer ranks the leading cause of gynecologic cancer-related death in the United States and the fifth most common cause of cancer-related mortality among American women. Increasing evidences have highlighted the vital role of macrophages M2/M1 proportion in tumor progression, prognosis and immunotherapy.

METHODS

Weighted gene co-expression network analysis (WGCNA) was performed to identify macrophages related markers. Integrative procedure including 10 machine learning algorithms were performed to develop a prognostic macrophage related signature (MRS) with TCGA, GSE14764, GSE140082 datasets. The role of MRS in tumor microenvironment (TME) and therapy response was evaluated with the data of CIBERSORT, MCPcounter, QUANTISEQ, XCELL, CIBERSORT-ABS, TIMER and EPIC, GSE91061 and IMvigor210 dataset.

RESULTS

The optimal MRS developed by the combination of CoxBoost and StepCox[forward] algorithm served as an independent risk factor in ovarian cancer. Compared with stage, grade and other established prognostic signatures, the current MRS had a better performance in predicting the overall survival rate of ovarian cancer patients. Low risk score indicated a higher TME score, higher level of immune cells, higher immunophenoscore, higher tumor mutational burden, lower TIDE score and lower IC50 value in ovarian cancer. The survival prediction nomogram had a good potential for clinical application in predicting the 1-, 3-, and 5-year overall survival rate of ovarian cancer patients.

CONCLUSION

All in all, the current study constructed a powerful prognostic MRS for ovarian cancer patients using 10 machine learning algorithms. This MRS could predict the prognosis and drug sensitivity in ovarian cancer.

Epididymal segment-specific miRNA and mRNA regulatory network at the single cell level

Spermatozoa released from the testis cannot fertilize an egg before becoming mature and motile in the epididymis. Based on three bulk and one single-cell RNA-seq (scRNA-seq) data series, we compared mRNA or miRNA expression between epididymal segment-specific samples and the other samples. Hereby, we identified 570 differentially expressed mRNAs (DE-mRNAs) and 23 differentially expressed miRNAs (DE-miRNAs) in the caput, 175 DE-mRNAs and 15 DE-miRNAs in the corpus, 946 DE-mRNAs and 12 DE-miRNAs in the cauda. In accordance with respective DE-miRNAs, we predicted upstream transcription factors (TFs) and downstream target genes. Subsequently, we intersected target genes of respective DE-miRNAs with corresponding DE-mRNAs, thereby obtaining 127 upregulated genes in the caput and 92 upregulated genes in cauda. Enriched upregulated pathways included cell motility-related pathways for the caput, smooth muscle-related pathways for the corpus, and immune-associated pathways for the cauda. Protein-protein interaction (PPI) network was constructed to extract key module for the caput and cauda, followed by identifying hub genes through cytohubba. Epididymis tissues from six mice were applied to validate hub genes expression using qRT-PCR, and 7 of the 10 genes displayed identical expression trends in mice caput/cauda. These hub genes were found to be predominantly distributed in spermatozoa using scRNA-seq data. In addition, target genes of DE-miRNAs were intersected with genes in the PPI network for each segment. Subsequently, the miRNA and mRNA regulatory networks for the caput and cauda were constructed. Conclusively, we uncover segment-specific miRNA-mRNA regulatory network, upstream TFs, and downstream pathways of the human epididymis, warranting further investigation into epididymal segment-specific functions.

Cellular and Molecular Mechanisms of the Tumor Stroma in Colorectal Cancer: Insights into Disease Progression and Therapeutic Targets

Colorectal cancer (CRC) is a major health burden worldwide and is the third most common type of cancer. The early detection and diagnosis of CRC is critical to improve patient outcomes. This review explores the intricate interplay between the tumor microenvironment, stromal interactions, and the progression and metastasis of colorectal cancer. The review begins by assessing the gut microbiome's influence on CRC development, emphasizing its association with gut-associated lymphoid tissue (GALT). The role of the Wnt signaling pathway in CRC tumor stroma is scrutinized, elucidating its impact on disease progression. Tumor budding, its effect on tumor stroma, and the implications for patient prognosis are investigated. The review also identifies conserved oncogenic signatures (COS) within CRC stroma and explores their potential as therapeutic targets. Lastly, the seed and soil hypothesis is employed to contextualize metastasis, accentuating the significance of both tumor cells and the surrounding stroma in metastatic propensity. This review highlights the intricate interdependence between CRC cells and their microenvironment, providing valuable insights into prospective therapeutic approaches targeting tumor-stroma interactions.

Introducing novel key genes and transcription factors associated with rectal cancer response to chemoradiation through co-expression network analysis

Preoperative radiochemotherapy is a promising therapeutic method for locally advanced rectal cancer patients. However, the response of colorectal cancer (CRC) patients to preoperative radiotherapy varies widely. In this study, we aimed to identify novel biomarkers that could predict the response of colorectal tumors to treatment using a systems biology approach. We applied the Weighted Gene Co-Expression Network Analysis to construct co-expression networks and evaluated the correlation of these networks with radiation using the module-trait relationship. We then identified hub genes and related transcription factors in the selected co-expression module. Our analysis of seven constructed modules revealed that one module, which contained 113 nodes and 6066 edges, had the strongest correlation with radiation effects on CRC (correlation = 0.85; p-value = 6e-7). By analyzing the selected module with the CytoHubba plugin, we identified four hub genes, including ZEB2, JAM2, NDN, and PPAP2A. We also identified seven important transcription factors, including KLF4, SUZ12, TCF4, NANOG, POU5F1, SOX2, and SMARCA4, which may play essential roles in regulating the four hub genes. In summary, our findings suggest that ZEB2, JAM2, NDN, and PPAP2A, along with the seven transcription factors related to these hub genes, may be associated with the response of colorectal tumors to chemoradiotherapy.