Identification of diagnostic genes in rheumatoid arthritis using integrated bioinformatics, machine learning, and experimental validation

利用整合生物信息学、机器学习和实验验证方法鉴定类风湿性关节炎的诊断基因

阅读:1

Abstract

BACKGROUND: Rheumatoid arthritis (RA) and osteoarthritis (OA) are prevalent joint diseases with overlapping clinical manifestations but distinct pathogenesis and treatment strategies. Misclassification may lead to inappropriate management. Therefore, accurate molecular discrimination between RA and OA is important. This study aimed to identify diagnostic genes associated with RA, with a particular emphasis on distinguishing RA from OA using integrated bioinformatics and machine learning approaches. METHODS: Public GEO transcriptomic datasets were analyzed to identify differentially expressed genes (DEGs) between the RA group and comparison groups. LASSO and SVM-RFE algorithms were applied for feature selection. Immune cell infiltration was estimated using the ssGSEA algorithm. A protein-protein interaction (PPI) network and transcription factor analysis were performed to explore potential regulatory mechanisms. Drug sensitivity analysis based on CellMiner IC50 data was conducted as an exploratory approach. In vitro validation was performed using TNF-α-stimulated HFLS-RA cells, followed by RT-qPCR analysis. RESULTS: Three key genes-EPYC, MAGED1, and LAP3-were identified as overlapping features between the LASSO and SVM-RFE models. ROC analysis demonstrated good discriminatory performance (AUC > 0.85). EPYC and LAP3 were associated with immune cell infiltration patterns. TNF-α stimulation significantly modulated the mRNA expression of these genes in the HFLS-RA cells. CONCLUSION: EPYC, MAGED1, and LAP3 are inflammation-associated genes with potential diagnostic relevance in RA. Further validation in larger independent cohorts and protein-level studies is needed to confirm their clinical applicability.

特别声明

1、本页面内容包含部分的内容是基于公开信息的合理引用;引用内容仅为补充信息,不代表本站立场。

2、若认为本页面引用内容涉及侵权,请及时与本站联系,我们将第一时间处理。

3、其他媒体/个人如需使用本页面原创内容,需注明“来源:[生知库]”并获得授权;使用引用内容的,需自行联系原作者获得许可。

4、投稿及合作请联系:info@biocloudy.com。