| 引用本文: | 康意,靳茜,周梦琪,郑慧娟,李丹文,王耀献,吕杰.基于机器学习筛选自噬相关糖尿病[]肾脏疾病的诊断基因及免疫浸润分析[J].生物信息学,2025,23(4):261-276. |
| KANG Yi,JIN Qian,ZHOU Mengqi,ZHENG Huijuan,LI Danwen,WANG Yaoxian,LÜ Jie.Screening of autophagy-related diagnostic genes in diabetic kidney disease using machine learning and immune infiltration analysis[J].Chinese Journal of Bioinformatics,2025,23(4):261-276. |
|
| |
|
|
| 本文已被：浏览 55次 下载 31次 |
码上扫一扫！ |
|
|
| 基于机器学习筛选自噬相关糖尿病[]肾脏疾病的诊断基因及免疫浸润分析 |
|
康意1,2,靳茜2,周梦琪3,郑慧娟1,李丹文1,2,王耀献1,吕杰1
|
|
(1.北京中医药大学东直门医院 北京 100700;2.北京中医药大学 北京 100029;3.北京市普仁医院,北京 100062)
|
|
| 摘要: |
| 为探讨自噬相关基因(ARGs)在糖尿病肾脏疾病(DKD)中的作用,通过机器学习方法筛选潜在的诊断基因,为DKD的早期诊断和治疗提供新的生物标志物和治疗靶点。从HADb数据库获取ARGs,对GSE96804数据集进行差异表达分析获得差异表达基因(DEGs),并与ARGs取交集后筛选出DKD-ARGs基因进行GO和KEGG富集分析。使用4种机器学习方法(LASSO、SVM、RF和BORUTA)筛选诊断基因并进行验证。通过无监督聚类分析和主成分分析识别自噬分子亚型并进行免疫浸润分析。获得1 526个DEGs(706个上调基因和820个下调基因),将DEGs与222个ARGs 取交集筛选出16个DKD-ARGs。GO和KEGG分析显示这些基因主要富集于细胞死亡、自噬、缺氧反应、炎症和免疫调节等生物过程和信号通路。机器学习算法最终筛选出3个关键诊断基因:CASP3、CDKN1B和PTEN,这些基因在验证集中显示出良好的预测性能。根据筛选出的基因进行一致性聚类获得Cluster1和Cluster2,免疫浸润分析显示两种自噬分子亚型之间存在显著的免疫细胞分布差异,揭示了自噬与免疫反应在DKD中的复杂关系。自噬相关诊断基因CASP3、CDKN1B和PTEN在DKD的早期诊断和治疗中具有重要应用潜力,为DKD提供了新的研究方向和治疗策略。 |
| 关键词: 糖尿病肾脏疾病 自噬 机器学习 免疫浸润 诊断基因 |
| DOI：10.12113/202407005 |
| 分类号:Q343.1+7 |
| 文献标识码:A |
| 基金项目:国家中医药管理局中医药传承与创新“百千万”人才工程项目(国中医药人教发[2018]12号)；中央高校基本科研业务费专项资金资助(2023-JYB-JBQN-020)；中华中医药学会联合攻关项目(2023DYPLHGG-11). |
|
| Screening of autophagy-related diagnostic genes in diabetic kidney disease using machine learning and immune infiltration analysis |
|
KANG Yi1,2,JIN Qian2,ZHOU Mengqi3,ZHENG Huijuan1,LI Danwen1,2,WANG Yaoxian1,LÜ Jie1
|
|
(1. Dongzhimen Hospital, Beijing University of Chinese Medicine, Beijing 100700,China; 2. Beijing University of Chinese Medicine, Beijing 100029,China; 3. Beijing Puren Hospital, Beijing 100062,China)
|
| Abstract: |
| To investigate the role of autophagy-related genes (ARGs) in diabetic kidney disease (DKD), screen potential diagnostic genes using machine learning methods, and thereby provide new biomarkers and targets for the early diagnosis and treatment of DKD. ARGs were obtained from the HADb database. Differential expression analysis of the GSE96804 dataset was performed to identify differentially expressed genes (DEGs), which were intersected with ARGs to screen DKD-ARGs genes for GO and KEGG enrichment analysis. Five machine learning methods (LASSO regression, SVM, random forest, XGBoost, and BORUTA) were used to screen and validate diagnostic genes. Unsupervised clustering analysis and principal component analysis were used to identify autophagy molecular subtypes and perform immune infiltration analysis. A total of 1526 DEGs (706 up-regulated genes and 820 down-regulated genes) were identified. The intersection of DEGs with 222 ARGs screened out 16 DKD-ARGs. GO and KEGG analyses showed that the genes were mainly enriched in biological processes and signaling pathways related to cell death, autophagy, hypoxic response, inflammation, and immune regulation. Machine learning algorithms ultimately identified three key diagnostic genes: CASP3, CDKN1B, and PTEN, which showed good predictive performance in the validation set.Consistency clustering based on the screened genes identified Cluster1 and Cluster2. Immune infiltration analysis showed significant differences in immune cell distribution between the two autophagy molecular subtypes, revealing the complex relationship between autophagy and immune response in DKD.The autophagy-related diagnostic genes CASP3, CDKN1B, and PTEN have significant potential for early diagnosis and treatment of DKD, providing new research directions and therapeutic strategies for DKD. |
| Key words: Diabetic kidney disease Autophagy Machine learning Immune infiltration Diagnostic genes |
|
|
|
|
