GO analysis results of the yellow module.

Parkinson’s disease is the second most common neurodegenerative disease in the world. We downloaded data on Parkinson’s disease and Ferroptosis-related genes from the GEO and FerrDb databases. We used WCGAN and Random Forest algorithm to screen out five Parkinson’s disease ferroptosis-related hub genes. Two genes were identified for the first time as possibly playing a role in Braak staging progression. Unsupervised clustering analysis based on hub genes yielded ferroptosis isoforms, and immune infiltration analysis indicated that these isoforms are associated with immune cells and may represent different immune patterns. FRHGs scores were obtained to quantify the level of ferroptosis modifications in each individual. In addition, differences in interleukin expression were found between the two ferroptosis subtypes. The biological functions involved in the hub gene are analyzed. The ceRNA regulatory network of hub genes was mapped. The disease classification diagnosis model and risk prediction model were also constructed by applying hub genes based on logistic regression. Multiple external datasets validated the hub gene and classification diagnostic model with some accuracy. This study explored hub genes associated with ferroptosis in Parkinson’s disease and their molecular patterns and immune signatures to provide new ideas for finding new targets for intervention and predictive biomarkers.

帕金森病（Parkinson’s disease）是全球范围内第二大高发的神经退行性疾病。本研究从GEO数据库与FerrDb数据库中下载了帕金森病及铁死亡（Ferroptosis）相关基因的数据集，采用WCGAN与随机森林算法筛选出5个与帕金森病铁死亡相关的枢纽基因（hub genes），其中2个基因为首次被证实可能参与帕金森病的Braak分期进展过程。基于枢纽基因开展的无监督聚类分析得到了铁死亡亚型，免疫浸润分析结果显示，这些亚型与免疫细胞密切相关且对应不同的免疫模式。我们构建了铁死亡相关枢纽基因（Ferroptosis-Related Hub Genes, FRHGs）评分以量化每位个体的铁死亡修饰水平；此外，两种铁死亡亚型之间的白细胞介素表达水平存在显著差异。本研究对枢纽基因所参与的生物学功能进行了系统分析，绘制了其内源竞争RNA（competing endogenous RNA, ceRNA）调控网络，并基于枢纽基因结合逻辑回归算法构建了疾病分类诊断模型与风险预测模型。多组外部数据集对枢纽基因及分类诊断模型进行了验证，结果显示二者具备一定的准确性。本研究探明了帕金森病中与铁死亡相关的枢纽基因及其分子模式与免疫特征，为发掘新型干预靶点与预测性生物标志物提供了全新的研究思路。