目的 探讨非酒精性脂肪性肝病(nonalcoholic fatty liver disease,NAFLD)的关键基因,构建NAFLD疾病进程的风险预测模型并进行验证.方法 基于GEO数据库中已有的NAFLD表达谱芯片数据集,采用差异表达基因分析、京都基因与基因组百科全书(Kyoto Encyclopedia of Genes and Genomes,KEGG)通路富集分析、蛋白质互作(protein-protein interaction,PPI)网络分析以及机器学习特征筛选方法,发现NAFLD疾病进程中的关键基因;同时,构建从单纯性脂肪肝(simple steatosis,SS)到非酒精性脂肪肝炎(nonalcoholic steatohepatitis,NASH)的NAFLD疾病进程风险预测模型;最后,通过体外分子生物学实验对关键基因进行验证.结果 GEO数据库中差异表达基因分析结果显示,正常对照(normal control,NC)组与NASH组比较有1247个差异表达基因,NC组与SS组之间有1088个差异表达基因,SS组与NASH组比较只有75个差异表达基因.结合前期研究和文献查阅以及KEGG通路富集分析结果显示,有4条与NAFLD疾病相关的共同信号通路,即胆固醇代谢(hsa04979)、半乳糖代谢(hsa00052)、PI3K/Akt信号通路(hsa04151)和PPAR信号通路(hsa03320)等.通过3种不同的机器学习特征筛选方法,最终得到4个共有基因,即AKR1B10、COL1A2、HKDC1、LAMC3.体外分子生物学实验结果显示,AKR1B10、COL1A2在NAFLD中均显著上调,与生物信息学分析结果一致.结论 本研究发现了 SS发展到NASH的关键基因,同时,构建了 NAFLD疾病进程的风险预测模型,为NAFLD疾病的有效控制、干预与临床诊断和治疗提供了有价值的方法与技术.
Discovery and Validation of Key Risk Genes in NAFLD Disease Progression
Objective To explore the key genes of nonalcoholic fatty liver disease(NAFLD),construct and verify the risk predic-tion model of NAFLD disease process Methods Based on the existing genomic microarray data for NAFLD from the GEO database,utili-zing differential expression gene analysis,Kyoto Encyclopedia of Genes and Genomes(KEGG)pathway enrichment analysis,protein-protein interaction(PPI)network analysis,and machine learning-based feature selection methods,key genes involved in the progression of NAFLD were identified,and a risk prediction model for the progression from simple steatosis(SS)to nonalcoholic steatohepatitis(NASH)was developed.Validation of these key genes was performed through molecular biology experiments in vitro.Results The re-sults of differential expression gene analysis in GEO database showed that there were 1247differential expression genes between the normal control(NC)group and the NASH group,1088differential expression genes between the NC group and the SS group,and 75differential expression genes between the SS group and NASH group.KEGG pathway enrichment analysis,informed by previous studies and litera-ture,identified 4 common signaling pathways related to NAFLD:cholesterol metabolism(hsa04979),galactose metabolism(hsa00052),PI3K/Akt signaling pathway(hsa04151),and PPAR signaling pathway(hsa03320).Using three distinct machine learning-based fea-ture selection methods,4 common genes were pinpointed:AKR1B10,COL1A2,HKDC1 and LAMC3.Molecular biology experiments in vitro showed that AKR1B10 and COL1A2 were significantly up-regulated in NAFLD,which was consistent with the bioinformatics analy-sis results.Conclusion This study identified key genes associated with the progression from SS to NASH and developed a risk prediction model for NAFLD disease progression.These findings offer valuable methods and technologies for the effective control,intervention,and clinical diagnosis and treatment of NAFLD.
万方数据
维普
