首页|基于加权基因共表达网络分析及机器学习构建铁死亡相关基因在多囊卵巢综合征的疾病模型及相关中药预测

基于加权基因共表达网络分析及机器学习构建铁死亡相关基因在多囊卵巢综合征的疾病模型及相关中药预测

扫码查看
目的:利用机器学习分析铁死亡基因与多囊卵巢综合征(PCOS)的相关性,构建诊断及预后风险模型,探讨PCOS的潜在铁死亡机制,并预测相关中药。方法:多芯片标准化合并GEO数据库的卵巢组织芯片,采用加权基因共表达网络分析(WGCNA)算法筛选与PCOS相关的铁死亡特征基因,对特征基因进行功能富集分析,利用多种机器学习进行构建并筛选疾病模型,提取最优的疾病特征基因,并构建风险模型,分析其与免疫浸润细胞及功能之间的相关性,对铁死亡基因进行中药及小分子化合物预测,并对小分子化合物与铁死亡靶点进行分子对接验证。结果:共提取26个铁死亡与PCOS的强相关交集靶基因,主要在铁死亡通路中的Xc-系统和铁代谢通路上发挥作用。通过WGCNA结合机器学习筛选以HMOX1、IFNA1、LPCAT3、FOXO4和PARP3为特征基因的RF模型有较优的模拟特性,并通过绘制列线图来预测各基因表达的患病风险性。上述模型基因与PCOS的发病呈正相关性。免疫浸润分析提示多种免疫细胞在两组之间存在显著浸润差异,预测的中药中丹参、三七可能是其潜在的分子药物来源,分子对接中丹酚酸B、葛根素与黄芪甲苷A的匹配程度最高。结论:以HMOX1、IFNA1、LPCAT3、FOXO4和PARP3为特征基因的铁死亡相关基因模型可为早期识别和治疗POCS提供新思路。铁死亡相关基因可通过干预多种途径在PCOS中发挥重要作用,中医药在干预PCOS中可能通过调节铁稳态代谢来介导铁死亡途径。
Construction of Disease Model for Ferroptosis-Related Genes in Polycystic Ovary Syndrome and Prediction of Related Traditional Chinese Medicine Based on Weighted Gene Co Expression Network Analysis and Machine Learning
Objective:To use machine learning to analyze the correlation between ferroptosis-related genes and polycystic ovary syndrome(PCOS),construct diagnostic and prognostic risk models,explore the potential ferroptosis mechanisms of PCOS,and predict related traditional Chinese medicine(TCM).Methods:Ovarian tissue microarray data from the GEO database were standardized and merged.The weighted gene co expression network analysis(WGCNA)algorithm was applied to select ferroptosis feature genes related to PCOS.Functional enrichment analysis was performed on these genes,and multiple machine learning methods were used to construct and select disease models.The optimal disease feature genes were extracted,and a risk model was constructed to analyze their correlation with immune infiltrating cells and functions.Ferroptosis-related genes were used for TCM and small molecule compound prediction,and molecular docking was conducted to validate the interactions between small molecule compounds and ferroptosis targets.Results:A total of 26 strongly correlated ferroptosis-related genes associated with PCOS were identified,primarily involved in the Xc-system and iron metabolism pathways offerroptosis.The RF model with HMOX1,IFNA1,LPCAT3.FOXO4,and PARP3 as feature genes screened through WGCNA combined with machine learning has better simulation characteristics,and the disease risk of each gene expression is predicted by drawing column charts.The aforementioned model genes showed a positive correlation with PCOS incidence.Immune infiltration analysis suggested significant differences in the infiltration of multiple immune cells between the two groups.The predicted TCM components,including Danshen(Salviae Miltiorrhizae Radix)and Sanqi(Notoginseng Radix),may be potential molecular sources of drugs,with the highest matching degree between salvianolic acid B,puerarin,and astragaloside A in molecular docking.Conclusion:The ferroptosis-related gene model with HMOX1,IFNA1,LPCAT3,FOXO4,and PARP3 as feature genes can provide new insights for the early identification and treatment of POCS.Ferroptosis-related genes play a crucial role in PCOS through intervention in various pathways.Traditional Chinese medicine may mediate the ferroptosis pathway by regulating iron homeostasis in the intervention of PCOS.

polycystic ovary syndromeferroptosisweighted gene co expression network analysismachine learningdisease modelsGEO databaseTCM prediction

唐昕燃、陈顺德、崔立生、胡真榕、沈浮、彭芮

展开 >

湖南医药学院附属永州市中医医院,湖南 永州 425000

多囊卵巢综合征 铁死亡 加权基因共表达网络分析 机器学习 疾病模型 GEO数据库 中药预测

永州市2022年度指导性科技计划项目

2022-YZKJZD-035

2024

中医药导报
湖南省中医药学会 湖南省中医管理局

中医药导报

CSTPCD
影响因子:0.952
ISSN:1672-951X
年,卷(期):2024.30(6)
  • 6