The combination of machine learning and untargeted metabolomics identifies the lipid metabolism -related gene CH25H as a potential biomarker in asthma

Ding, Xuexuan; Qin, Jingtong; Huang, Fangfang; Feng, Fuhai; Luo, Lianxiang

doi:10.1007/s00011-023-01732-0

Cited by 8 publications

(3 citation statements)

References 67 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…ANXA2 is filtered by machine learning algorithms. A previous study of ours also demonstrated the robustness of the machine learning results ( 36 ). In this study, the differential expression of ANXA2 in the three datasets represents its reliability.…”

Section: Discussionsupporting

confidence: 72%

Machine learning identifies ferroptosis-related gene ANXA2 as potential diagnostic biomarkers for NAFLD

Qin,

Cao,

Ding

et al. 2023

Front. Endocrinol.

Self Cite

View full text Add to dashboard Cite

IntroductionNon-alcoholic fatty liver disease (NAFLD), a major cause of chronic liver disease, still lacks effective therapeutic targets today. Ferroptosis, a type of cell death characterized by lipid peroxidation, has been linked to NAFLD in certain preclinical trials, yet the exact molecular mechanism remains unclear. Thus, we analyzed the relationship between ferroptosis genes and NAFLD using high-throughput data.MethodWe utilized a total of 282 samples from five datasets, including two mouse ones, one human one, one single nucleus dataset and one single cell dataset from Gene Expression Omnibus (GEO), as the data basis of our study. To filter robust treatment targets, we employed four machine learning methods (LASSO, SVM, RF and Boruta). In addition, we used an unsupervised consensus clustering algorithm to establish a typing scheme for NAFLD based on the expression of ferroptosis related genes (FRGs). Our study is also the first to investigate the dynamics of FRGs throughout the disease process by time series analysis. Finally, we validated the relationship between core gene and ferroptosis by in vitro experiments on HepG2 cells.ResultsWe discovered ANXA2 as a central focus in NAFLD and indicated its potential to boost ferroptosis in HepG2 cells. Additionally, based on the results obtained from time series analysis, ANXA2 was observed to significantly define the disease course of NAFLD. Our results demonstrate that implementing a ferroptosis-based staging method may hold promise for the diagnosis and treatment of NAFLD.ConclusionOur findings suggest that ANXA2 may be a useful biomarker for the diagnosis and characterization of NAFLD.

show abstract

Section: Discussionsupporting

confidence: 72%

Machine learning identifies ferroptosis-related gene ANXA2 as potential diagnostic biomarkers for NAFLD

Qin,

Cao,

Ding

et al. 2023

Front. Endocrinol.

Self Cite

View full text Add to dashboard Cite

show abstract

“…In their study, Ding et al employed the WGCNA technique along with five machine learning algorithms to identify hub genes associated with lipid metabolism. Their findings indicated that CH25H exhibited potential as a biomarker for asthma in relation to lipid metabolism (Ding et al 2023 ). Accordingly, the inadequacy of a singular biomarker for this diverse disease was apparent.…”

Section: Discussionmentioning

confidence: 99%

CDC167 exhibits potential as a biomarker for airway inflammation in asthma

Zhong,

Wu,

Cai

et al. 2024

Mamm Genome

View full text Add to dashboard Cite

Current asthma treatments have been discovered to decrease the risk of disease progression. Herein, we aimed to characterize novel potential therapeutic targets for asthma. Differentially expressed genes (DEGs) for GSE64913 and GSE137268 datasets were characterized. Weighted correlation network analysis (WGCNA) was used to identify trait-related module genes within the GSE67472 dataset. The intersection of the module genes of interest, as well as the DEGs, comprised the key module genes that underwent additional candidate gene screening using machine learning. In addition, a bioinformatics-based approach was used to analyze the relative expression levels, diagnostic values, and reverently enriched pathways of the screened candidate genes. Furthermore, the candidate genes were silenced in asthmatic mice, and the inflammation and lung injury in the mice were validated. A total of 1710 DEGs were characterized in GSE64913 and GSE137268 for asthma patients. WGCNA identified 2367 asthma module genes, of which 285 overlapped with 1710 DEGs. Four candidate genes, CDC167, POSTN, SEC14L1, and SERPINB2, were validated using the intersection genes of three machine learning algorithms, including Least Absolute Shrinkage and Selection Operator, Random Forest, and Support Vector Machine. All the candidate genes were significantly upregulated in asthma patients and demonstrated diagnostic utility for asthma. Furthermore, silencing CDC167 reduced the levels of inflammatory cytokines significantly and alleviated lung injury in ovalbumin (OVA)-induced asthmatic mice. Our study demonstrated that CDC167 exhibits potential as diagnostic markers and therapeutic targets for asthma patients.

show abstract

“…The SVM-RFE algorithm is one of the most supervised machine learning methods to handle high dimensional data [34]. The Boruta algorithm is a supervised classification method utilized for feature selection, aiming to ascertain features associated with a given classification task [35]. These methods are widely used to identify biomarkers with superior accuracy and good interpretability [27][28][29].…”

Section: Identification Of Diagnostic Biomarkers and Construction Of ...mentioning

confidence: 99%

Integrative Analysis of Machine Learning and Molecule Docking Simulations for Ischemic Stroke Diagnosis and Therapy

Song,

Zaidi,

et al. 2023

Molecules

View full text Add to dashboard Cite

Due to the narrow therapeutic window and high mortality of ischemic stroke, it is of great significance to investigate its diagnosis and therapy. We employed weighted gene coexpression network analysis (WGCNA) to ascertain gene modules related to stroke and used the maSigPro R package to seek the time-dependent genes in the progression of stroke. Three machine learning algorithms were further employed to identify the feature genes of stroke. A nomogram model was built and applied to evaluate the stroke patients. We analyzed single-cell RNA sequencing (scRNA-seq) data to discern microglia subclusters in ischemic stroke. The RNA velocity, pseudo time, and gene set enrichment analysis (GSEA) were performed to investigate the relationship of microglia subclusters. Connectivity map (CMap) analysis and molecule docking were used to screen a therapeutic agent for stroke. A nomogram model based on the feature genes showed a clinical net benefit and enabled an accurate evaluation of stroke patients. The RNA velocity and pseudo time analysis showed that microglia subcluster 0 would develop toward subcluster 2 within 24 h from stroke onset. The GSEA showed that the function of microglia subcluster 0 was opposite to that of subcluster 2. AZ_628, which screened from CMap analysis, was found to have lower binding energy with Mmp12, Lgals3, Fam20c, Capg, Pkm2, Sdc4, and Itga5 in microglia subcluster 2 and maybe a therapeutic agent for the poor development of microglia subcluster 2 after stroke. Our study presents a nomogram model for stroke diagnosis and provides a potential molecule agent for stroke therapy.

show abstract

The combination of machine learning and untargeted metabolomics identifies the lipid metabolism -related gene CH25H as a potential biomarker in asthma

Cited by 8 publications

References 67 publications

Machine learning identifies ferroptosis-related gene ANXA2 as potential diagnostic biomarkers for NAFLD

Machine learning identifies ferroptosis-related gene ANXA2 as potential diagnostic biomarkers for NAFLD

CDC167 exhibits potential as a biomarker for airway inflammation in asthma

Integrative Analysis of Machine Learning and Molecule Docking Simulations for Ischemic Stroke Diagnosis and Therapy

Contact Info

Product

Resources

About