Sun Hongyu, Yang Jin, Li Xiaohui, Lyu Yi, Xu Zhaomeng, He Hui, Tong Xiaomin, Ji Tingyu, Ding Shihan, Zhou Chaoli, Han Pengyong, Zheng Jinping
Department of Health Toxicology, School of Public Health in Shanxi Medical University, Taiyuan, China.
The Central Lab, Changzhi Medical College, Changzhi, China.
Front Comput Neurosci. 2022 Sep 21;16:1001546. doi: 10.3389/fncom.2022.1001546. eCollection 2022.
While Alzheimer's disease (AD) can cause a severe economic burden, the specific pathogenesis involved is yet to be elucidated. To identify feature genes associated with AD, we downloaded data from three GEO databases: GSE122063, GSE15222, and GSE138260. In the filtering, we used AD for search keywords, Homo sapiens for species selection, and established a sample size of > 20 for each data set, and each data set contains Including the normal group and AD group. The datasets GSE15222 and GSE138260 were combined as a training group to build a model, and GSE122063 was used as a test group to verify the model's accuracy. The genes with differential expression found in the combined datasets were used for analysis through Gene Ontology (GO) and The Kyoto Encyclopedia of Genes and Genome Pathways (KEGG). Then, AD-related module genes were identified using the combined dataset through a weighted gene co-expression network analysis (WGCNA). Both the differential and AD-related module genes were intersected to obtain AD key genes. These genes were first filtered through LASSO regression and then AD-related feature genes were obtained for subsequent immune-related analysis. A comprehensive analysis of three AD-related datasets in the GEO database revealed 111 common differential AD genes. In the GO analysis, the more prominent terms were cognition and learning or memory. The KEGG analysis showed that these differential genes were enriched not only in In the KEGG analysis, but also in three other pathways: neuroactive ligand-receptor interaction, cAMP signaling pathway, and Calcium signaling pathway. Three AD-related feature genes (SST, MLIP, HSPB3) were finally identified. The area under the ROC curve of these AD-related feature genes was greater than 0.7 in both the training and the test groups. Finally, an immune-related analysis of these genes was performed. The finding of AD-related feature genes (SST, MLIP, HSPB3) could help predict the onset and progression of the disease. Overall, our study may provide significant guidance for further exploration of potential biomarkers for the diagnosis and prediction of AD.
虽然阿尔茨海默病（AD）会造成严重的经济负担，但其具体发病机制尚待阐明。为了识别与AD相关的特征基因，我们从三个基因表达综合数据库（GEO）下载了数据：GSE122063、GSE15222和GSE138260。在筛选过程中，我们使用AD作为搜索关键词，选择智人为物种，并且为每个数据集设定样本量大于20，每个数据集都包含正常组和AD组。将数据集GSE15222和GSE138260合并为训练组来构建模型，使用GSE122063作为测试组来验证模型的准确性。在合并数据集中发现的差异表达基因通过基因本体论（GO）和京都基因与基因组百科全书通路（KEGG）进行分析。然后，通过加权基因共表达网络分析（WGCNA）使用合并数据集识别AD相关模块基因。将差异基因和AD相关模块基因进行交集运算以获得AD关键基因。这些基因首先通过套索回归进行筛选，然后获得AD相关特征基因用于后续的免疫相关分析。对GEO数据库中三个与AD相关的数据集进行综合分析，揭示了111个常见的差异AD基因。在GO分析中，较为突出的术语是认知以及学习或记忆。KEGG分析表明，这些差异基因不仅在KEGG分析中富集，还在其他三个通路中富集：神经活性配体-受体相互作用、环磷酸腺苷（cAMP）信号通路和钙信号通路。最终鉴定出三个与AD相关的特征基因（促生长抑素[SST]、肌肉 LIM 蛋白[MLIP]、热休克蛋白β3 [HSPB3]）。这些与AD相关的特征基因在训练组和测试组中的曲线下面积（ROC）均大于0.7。最后，对这些基因进行了免疫相关分析。发现与AD相关的特征基因（SST、MLIP、HSPB3）有助于预测疾病的发病和进展。总体而言，我们的研究可能为进一步探索用于AD诊断和预测的潜在生物标志物提供重要指导。
