判别式
计算生物学
疾病
转录组
诊断生物标志物
基因
Lasso(编程语言)
生物信息学
候选基因
基因表达
生物标志物
生物
生物标志物发现
医学
基因表达谱
机器学习
病态的
回归
基因调控网络
相关性(法律)
生物途径
小RNA
作者
Mingchen Zhao,Xinlu Sui,J S Li,Q Wang,Y M Sun,Haojia Zheng,Bin Li
标识
DOI:10.3389/fnagi.2026.1857642
摘要
Background: Alzheimer's disease (AD) lacks reliable early diagnostic biomarkers and effective disease-modifying therapies. Ferroptosis has been increasingly implicated in AD pathogenesis; however, ferroptosis-related genes with diagnostic and mechanistic relevance remain insufficiently characterized. Methods: Transcriptomic datasets were obtained from the GEO database and analyzed using differential expression analysis and WGCNA, followed by integration with ferroptosis-related gene sets. Key targets were identified through PPI networks and machine learning approaches, including LASSO regression and random forest. A diagnostic model was constructed and its discriminative performance was assessed across independent datasets by cohort-specific refitting. Functional roles were explored using GSEA, and experimental validation was conducted by qPCR in Aβ-treated PC12 cells. Potential therapeutic compounds were predicted using the CMAP database. Results: , were consistently selected as core features by machine learning algorithms. A diagnostic model based on these genes demonstrated strong performance in the training cohort (AUC > 0.8) and reproducible discriminative ability across external datasets, albeit with cohort-dependent variability. Functional analysis suggested that these genes are involved in pathways related to neuroinflammation, lysosomal activity, and oxidative stress, indicating a potential link between ferroptosis and key pathological processes in AD. qPCR validation confirmed the differential expression trends of these genes. In addition, several candidate compounds, including etomoxir and tubastatin A, were predicted to potentially modulate these pathways. Conclusion: , were identified in Alzheimer's disease and showed consistent discriminative ability across multiple cohorts, with links to neuroinflammation, lysosomal function, and oxidative stress. These findings highlight their potential as candidate biomarkers warranting further investigation, although prospective validation in independent cohorts is required to confirm their diagnostic utility.
科研通智能强力驱动
Strongly Powered by AbleSci AI