基于骨关节炎端粒相关基因的人工神经网络模型的构建OA
Construction of artificial neural network model based on telomere related genes in osteoarthritis
建立基于 GEO 数据库骨关节炎端粒相关基因的随机森林和人工神经网络(Artificial neural network,ANN)联合诊断模型并评价其效果.通过 GEO 数据库获取两份骨关节炎芯片 GSE55235 和 GSE129147 合并作为训练组数据集,并进行 GO 和KEGG 通路富集分析,探讨其潜在生物学功能.随后,使用随机森林算法筛选骨关节炎端粒相关基因,并构建用于诊断的ANN 模型.使用外部数据集 GSE98918 对模型进行验证,通过曲线下面积(Area under the curve,AUC)分析评价 ANN 模型的临床应用价值.利用 CIBERSORT 算法预估骨关节炎与筛选出的潜在生物标志物的生物信息学关联.最后进行关键基因与免疫细胞浸润相关性分析.共得到 696 个(上调 440 个,下调 256 个)端粒相关差异基因.GO 和 KEGG 通路富集分析涉及多个生物学功能.进一步通过随机森林筛选得到4 个关键基因(TREM2、ADCY2、MSR1、TWF2),并构建了ANN 诊断模型.使用该模型绘制了训练组和验证组诊断的受试者操作特征(Receiver operating characteristic,ROC)曲线,训练组 AUC 值为 0.999.验证组 AUC 为 0.826.免疫细胞浸润分析显示记忆静息 T 细胞 CD4 和静息肥大细胞在正常组与 OA 组之间存在明显差异.各免疫细胞正相关较高的幼稚 T 细胞 CD4 与浆细胞;负相关较高的为单核细胞与巨噬细胞 M0.相关性分析结果显示关键基因与多种免疫细胞的表达存在显著相关性.本研究构建了骨关节炎端粒相关基因的 ANN 诊断模型,可用于临床诊断骨关节炎,为探索骨关节炎发病机制提供了一个新视角.
To establish a joint random forest and artificial neural network(ANN)diagnosis model of telomere related genes in osteoarthritis based on GEO database and evaluate its effect.Two osteoarthritis chips GSE55235 and GSE129147 were obtained from GEO database as training group data set,and GO and KEGG pathway enrichment analysis was performed to explore their potential biological functions.Subsequently,a random forest algorithm was used to screen telomere related genes for osteoarthritis,and an ANN model was constructed for diagnosis.The external data set GSE98918 was used to verify the model,and the clinical application value of ANN model was evaluated by area under the curve(AUC)analysis.CIBERSORT algorithm was used to estimate the bioinformatics association between osteoarthritis and potential biomarkers.Finally,a correlation analysis between key genes and immune cell infiltration was conducted.A total of 696 telomere related differential genes(440 up-regulated and 256 down-regulated)were identified.Enrichment analysis of GO and KEGG pathways involves multiple biological functions.4 key genes(TREM2,ADCY2,MSR1,TWF2)were identified by random forest screening,and an ANN diagnostic model was constructed.The receiver Operating characteristic(ROC)curves for the diagnosis of the training group and the validation group were plotted using this model,and the AUC value of the training group was 0.999.The AUC of the verification group was 0.826.Immune cell infiltration analysis showed significant differences in T-cell CD4 memory resting and resting mast cells between normal and OA groups.Each immune cell positively correlated with higher naive T cell CD4 and plasma cell;The higher negative correlation was found between monocyte and macrophage M0.The correlation analysis results showed a significant correlation between key genes and the expression of various immune cells.The ANN diagnostic model of telomere related genes in osteoarthritis was constructed,which can be used for clinical diagnosis of osteoarthritis,and provides a new perspective for exploring the pathogenesis of osteoarthritis.
郑欣欣;郑福增
河南中医药大学 骨伤学院,郑州 450008河南省中医院,郑州 450002
医药卫生
骨关节炎端粒人工神经网络随机森林免疫细胞浸润
OsteoarthritisTelomereArtificial neural networkRandom forestImmune cell infiltration
《生物信息学》 2026 (3)
253-262,10
河南省中医药科学研究专项课题(No.2024ZY1012、2022ZYZD08).
评论