Construction and evaluation of an integrated predictive model for chronic kidney disease based on the random forest and artificial neural network approaches

Molecular Subtype Identification and Potential Drug Prediction Based on Anoikis-Related Genes Expression in Keratoconus

Purpose

Anoikis is a special apoptosis accompanied by the loss of extracellular matrix (ECM) environment and the decomposition of ECM is an important process in the occurrence of keratoconus (KC). This study aims to describe the expression profile of anoikis-related genes (ARGs) in KC samples, identify differentially expressed genes (DEGs), characterize the biological functions and immune characteristics of different molecular subtypes of KC and predict potential drugs based on the construction of a co-expression network.

Methods

First, we identified molecular subtypes by optimal clustering K based on the expression profile of ARGs in the KC dataset and analyzed the differences of functional and immune characteristics. Then a weighted gene co-expression network was constructed based on cluster analysis to obtain hub genes and protein-protein interaction network was constructed to analyze hub nodes and predict potential node-targeting drugs.

Results

By comparing the expression profile between disease and normal samples, we found that there were significant differences in ARGs such as BCL2, CAV1, and CEACAM5. Consistent cluster analysis identified two definite clusters on the basis of ARGs expression difference. Kyoto Encyclopedia of Genes and Genomes and Gene Ontology enrichment analysis showed that DEGs were enriched significantly in pathways like ECM receptor interaction, chemokine signal, notch signal, focal adhesion, and functional sets like proteolysis, anoikis, regulation of natural killer and T-cell proliferation. CIBERSORT calculation showed that there were significant differences between the two subtypes on immune cell infiltration (monocytes and plasma) and immune molecules (CCL11, CCL14, HLA-A, HLA-B, and so on). Then, co-expression network was constructed based on cluster phenotype, 5202 genes were selected as hub genes, and 321 HubDEGs were obtained after intersection with significant DEGs. Seven hub nodes, EIF4G1, KHSRP, PABPC1, POLR2A, PTBP1, RPS19, and SMARCA4, were identified and matched drugs or small molecular compounds. Insulin and dexamethasone were identified as potential target drugs.

Conclusions

We revealed the differential expression of ARGs in KC samples, and identified two distinct subtypes that showed significant differences in biological function and immune infiltration. The identification of hub gene nodes elucidated their therapeutic value on predicted potential drugs.

Establishment of a machine learning predictive model for non-alcoholic fatty liver disease: A longitudinal cohort study

BACKGROUND AND AIMS

Non-alcoholic fatty liver disease (NAFLD) is a common chronic liver disease, which lacks effective drug treatments. This study aimed to construct an eXtreme Gradient Boosting (XGBoost) prediction model to identify or evaluate potential NAFLD patients.

METHODS AND RESULTS

We conducted a longitudinal study of 22,140 individuals from the Beijing Health Management Cohort. Variable filtering was performed using the least absolute shrinkage and selection operator. Random Over Sampling Examples was used to address imbalanced data. Next, the XGBoost model and the other three machine learning (ML) models were built using balanced data. Finally, the variable importance of the XGBoost model was ranked. Among four ML algorithms, we got that the XGBoost model outperformed the other models with the following results: accuracy of 0.835, sensitivity of 0.835, specificity of 0.834, Youden index of 0.669, precision of 0.831, recall of 0.835, F-1 score of 0.833, and an area under the curve of 0.914. The top five variables with the greatest impact on the onset of NAFLD were aspartate aminotransferase, cardiometabolic index, body mass index, alanine aminotransferase, and triglyceride-glucose index.

CONCLUSION

The predictive model based on the XGBoost algorithm enables early prediction of the onset of NAFLD. Additionally, assessing variable importance provides valuable insights into the prevention and treatment of NAFLD.

Cell Profiling of Acute Kidney Injury to Chronic Kidney Disease Reveals Novel Oxidative Stress Characteristics in the Failed Repair of Proximal Tubule Cells

Chronic kidney disease (CKD) is a major public health issue around the world. A significant number of CKD patients originates from acute kidney injury (AKI) patients, namely "AKI-CKD". CKD is significantly related to the consequences of AKI. Damaged renal proximal tubular (PT) cell repair has been widely confirmed to indicate the renal prognosis of AKI. Oxidative stress is a key damage-associated factor and plays a significant role throughout the development of AKI and CKD. However, the relationships between AKI-CKD progression and oxidative stress are not totally clear and the underlying mechanisms in "AKI-CKD" remain indistinct. In this research, we constructed unilateral ischemia-reperfusion injury (UIRI)-model mice and performed single-nucleus RNA sequencing (snRNA-seq) of the kidney samples from UIRI and sham mice. We obtained our snRNA-seq data and validated the findings based on the joint analysis of public databases, as well as a series of fundamental experiments. Proximal tubular cells associated with failed repair express more complete senescence and oxidative stress characteristics compared to other subgroups. Furthermore, oxidative stress-related transcription factors, including Stat3 and Dnmt3a, are significantly more active under the circumstance of failed repair. What is more, we identified abnormally active intercellular communication between PT cells associated with failed repair and macrophages through the APP-CD74 pathway. More notably, we observed that the significantly increased expression of CD74 in hypoxia-treated TECs (tubular epithelial cells) was dependent on adjacently infiltrated macrophages, which was essential for the further deterioration of failed repair in PT cells. This research provides a novel understanding of the process of AKI to CKD progression, and the oxidative stress-related characteristics that we identified might represent a potentially novel therapeutic strategy against AKI.

Role of anoikis-related gene PLK1 in kidney renal papillary cell carcinoma: a bioinformatics analysis and preliminary verification on promoting proliferation and migration

Background: Kidney renal papillary cell carcinoma (KIRP) is a rare malignancy with a very poor prognosis. Anoikis is a specific form of apoptosis involved in carcinogenesis, but the role of anoikis in KIRP has not been explored. Methods: Anoikis-related genes (ARGs) were obtained from the GeneCards database and Harmonizome database and were used to identify different subtypes of KIRP and construct a prognostic model of KIRP. In addition, we also explored the immune microenvironment and enrichment pathways among different subtypes by consensus clustering into different subtypes. Drug sensitivity analysis was used to screen for potential drugs. Finally, we verified the mRNA and protein expression of the independent prognostic gene PLK1 in patient tissues and various cells and further verified the changes in relevant prognostic functions after constructing a PLK1 stable knockdown model using ShRNA. Results: We identified 99 differentially expressed anoikis-related genes (DEGs) associated with KIRP survival, and selected 3 genes from them to construct a prognostic model, which can well predict the prognosis of KIRP patients. Consensus clustering divided KIRP into two subtypes, and there was a significant difference in survival rates between the two subtypes. Immune profiling revealed differing immune statuses between the two subtypes, and functional analysis reveals the differential activity of different functions in different subtypes. Drug sensitivity analysis screened out 15 highly sensitive drugs in the high-risk group and 11 highly sensitive drugs in the low-risk group. Univariate and multivariate Cox regression analysis confirmed that PLK1 was an independent prognostic factor in KIRP, and its mRNA and protein expression levels were consistent with gene differential expression levels, both of which were highly expressed in KIRP. Functional verification of PLK1 in KIRP revealed significant results. Specifically, silencing PLK1 inhibited cell proliferation, clonogenicity, and migration, which indicated that PLK1 plays an important role in the proliferation and migration of KIRP. Conclusion: The prognosis model constructed by ARGs in this study can accurately predict the prognosis of KIRP patients. ARGs, especially PLK1, play an important role in the development of KIRP. This research can help doctors provide individualized treatment plans for KIRP patients and provide researchers with new research ideas.

Identification of PDK4 as Hub Gene for Diabetic Nephropathy Using Co-Expression Network Analysis

INTRODUCTION

Diabetic nephropathy (DN) is related to type 1 and type 2 diabetes. They are the leading cause of end-stage renal disease, but the underling specific pathogenesis of DN is not yet clear. Our study was conducted to explore how DN changed the transcriptome profiles in the kidney.

METHODS

The gene expression profile of microdissected glomeruli of 41 type 2 DN patients and 20 healthy controls were included. The sample dataset GSE96804 was obtained from the GEO database. Differentially expressed genes (DEGs) were analyzed in R with the limma package and the important modules were found by weighted gene co-expression network analysis (WGCNA) clustering. The modules were then analyzed based on Gene Ontology (GO) gene set enrichment analysis, and the hub genes were found out. We next validated the hub gene, PDK4, in a cell model of DN. We also constructed the PDK4-related PPI network to investigate the correlation between PDK4 expression and other genes.

RESULTS

Heatmap and volcano map were drawn to illustrate the mRNA expression profile of 1,204 DEGs in both samples of DN patients and the control group. Using WGCNA, we selected the blue module in which genes showed the strongest correlation with the phenotype and the smallest p value. We also identified PDK4 as a hub gene. PDK4 expression was upregulated in human diabetic kidney tissue. Moreover, PDK4 was speculated to play a role in glomerular basement membrane development and kidney development according to the enrichment of functions and signaling pathways. Furthermore, PDK4 and two key genes GSTA2 and G6PC protein expression were verified highly expressed in the cell model of DN.

CONCLUSION

During the pathogenesis of DN, many genes may change expression in a coordinated manner. The discovery of PDK4 as key gene using WGCNA is of great significance for the development of new treatment strategies to block the development of DN.

A united model for diagnosing pulmonary tuberculosis with random forest and artificial neural network

Background: Pulmonary tuberculosis (PTB) is a chronic infectious disease and is the most common type of TB. Although the sputum smear test is a gold standard for diagnosing PTB, the method has numerous limitations, including low sensitivity, low specificity, and insufficient samples.

Methods: The present study aimed to identify specific biomarkers of PTB and construct a model for diagnosing PTB by combining random forest (RF) and artificial neural network (ANN) algorithms. Two publicly available cohorts of TB, namely, the GSE83456 (training) and GSE42834 (validation) cohorts, were retrieved from the Gene Expression Omnibus (GEO) database. A total of 45 and 61 differentially expressed genes (DEGs) were identified between the PTB and control samples, respectively, by screening the GSE83456 cohort. An RF classifier was used for identifying specific biomarkers, following which an ANN-based classification model was constructed for identifying PTB samples. The accuracy of the ANN model was validated using the receiver operating characteristic (ROC) curve. The proportion of 22 types of immunocytes in the PTB samples was measured using the CIBERSORT algorithm, and the correlations between the immunocytes were determined.

Results: Differential analysis revealed that 11 and 22 DEGs were upregulated and downregulated, respectively, and 11 biomarkers specific to PTB were identified by the RF classifier. The weights of these biomarkers were determined and an ANN-based classification model was subsequently constructed. The model exhibited outstanding performance, as revealed by the area under the curve (AUC), which was 1.000 for the training cohort. The AUC of the validation cohort was 0.946, which further confirmed the accuracy of the model.

Conclusion: Altogether, the present study successfully identified specific genetic biomarkers of PTB and constructed a highly accurate model for the diagnosis of PTB based on blood samples. The model developed herein can serve as a reliable reference for the early detection of PTB and provide novel perspectives into the pathogenesis of PTB.

ILSHIP: An interpretable and predictive model for hypothyroidism

Hypothyroidism is one of the common endocrine diseases, and its incidence is increasing year by year. Due to the insidious nature of this disease, it often leads to delayed treatment and even misdiagnosis. This paper proposes ILSHIP, an interpretable predictive model for hypothyroidism, to reduce its diagnostic complexity as well as improve the predictive performance and interpretability of existing models. First, the ILSHIP prediction model was built based on label encoding, missing value processing, feature selection, and data enhancement of the dataset. Second, the comprehensive performance of ILSHIP was compared with twelve existing related study models and eleven mainstream models, such as XGBoost and MLP. The experimental results showed that, based on the optimal hyperparameters the ILSHIP model can achieve 99.392%, 99.437%, 99.348%, 99.381%, and 99.960% in accuracy, recall, specificity, F1, and AUC, respectively. The accuracy of the ILSHIP model was about 0.7%-15.4% higher than the existing models. By introducing the SHAP framework into the ILSHIP model, important features affecting hypothyroidism such as thyroid stimulating hormone (TSH) and free thyroxine index (FTI) were also identified, and the influencing factors for different individuals were finally analyzed to provide a basis for medical personnel to monitor the condition.

Identification of biomarkers for the diagnosis of chronic kidney disease (CKD) with non-alcoholic fatty liver disease (NAFLD) by bioinformatics analysis and machine learning

BACKGROUND

Chronic kidney disease (CKD) and non-alcoholic fatty liver disease (NAFLD) are closely related to immune and inflammatory pathways. This study aimed to explore the diagnostic markers for CKD patients with NAFLD.

METHODS

CKD and NAFLD microarray data sets were screened from the GEO database and analyzed the differentially expressed genes (DEGs) in GSE10495 of CKD date set. Weighted Gene Co-Expression Network Analysis (WGCNA) method was used to construct gene coexpression networks and identify functional modules of NAFLD in GSE89632 date set. Then obtaining NAFLD-related share genes by intersecting DEGs of CKD and modular genes of NAFLD. Then functional enrichment analysis of NAFLD-related share genes was performed. The NAFLD-related hub genes come from intersection of cytoscape software and machine learning. ROC curves were used to examine the diagnostic value of NAFLD related hub genes in the CKD data sets and GSE89632 date set of NAFLD. CIBERSORTx was also used to explore the immune landscape in GSE104954, and the correlation between immune infiltration and hub genes expression was investigated.

RESULTS

A total of 45 NAFLD-related share genes were obtained, and 4 were NAFLD-related hub genes. Enrichment analysis showed that the NAFLD-related share genes were significantly enriched in immune-related pathways, programmed cell death, and inflammatory response. ROC curve confirmed 4 NAFLD-related hub genes in CKD training set GSE104954 and other validation sets. Then they were used as diagnostic markers for CKD. Interestingly, these 4 diagnostic markers of CKD also showed good diagnostic value in the NAFLD date set GSE89632, so these genes may be important targets of NAFLD in the development of CKD. The expression levels of the 4 diagnostic markers for CKD were significantly correlated with the infiltration of immune cells.

CONCLUSION

4 NAFLD-related genes (DUSP1, NR4A1, FOSB, ZFP36) were identified as diagnostic markers in CKD patients with NAFLD. Our study may provide diagnostic markers and therapeutic targets for CKD patients with NAFLD.

Identification of Gene Coexpression Modules and Prognostic Genes Associated with Papillary Thyroid Cancer

Thyroid cancer is a great part of the endocrine tumor with an increasing incidence. Papillary thyroid carcinoma (PTC) is the most common subtype. With the enormous pace taken in the microarray technology, bioinformatics is applied in data mining more frequently. Weighted gene coexpression network analysis (WGCNA) can perform analysis combining clinic information. We performed WGCNA for prognostic genes associated with PTC. From the GEO profile, we got ten modules. We identified a key module that was closest to patients’ survival time. Then, we screened five hub genes (ATRX, BOD1L1, CEP290, DCAF16, and NEK1) from the key module based on the clinical information from TCGA. These five genes not only significantly differ between the normal and tumor groups but have prognostic value. The receiver operating characteristic (ROC) curve indicated that they had the potential to serve as prognostic genes. We performed next-generation sequencing using the PTC tissue to get more convincing evidence. Besides, we established a new signature and verified it through K-M plots and ROC. The signature could be an independent factor for the prognosis of PTC, and we built a nomogram to carry out a quantitative study. In a word, the hub genes we explored in the study deserved more basic and clinical research.