WGCNA联合机器学习筛选前列腺癌的关键基因及潜在治疗中药的预测OA
Screening of Key Genes in Prostate Cancer and Prediction of Potential Therapeutic Traditional Chinese Medicines by Combining WGCNA and Machine Learning
目的 筛选前列腺癌的关键基因,并预测具有治疗作用的中药和化合物.方法 从TCGA数据库和GEO数据库中下载数据集GSE26910、GSE55945.使用R软件分析进行差异分析、加权基因共表达网络分析、LASSO回归分析和SVM-RFE分析筛选前列腺癌的核心基因,并进行GO功能富集分析和KEGG通路富集分析,最后在Coremine Medical和TCMSP数据库中进行中药和化合物分析,并利用分子对接进行一步验证.结果 最终筛选得到691个DEGs,主要富集于肌肉收缩、肌肉系统过程、循环系统中的血管过程、间充质发育、平滑肌收缩、调节血液循环生物学过程和通路.根据WGCNA分析得到155个关键模块基因,与DEGs交集得到138个枢纽基因,基于两种机器学习算法LASSO回归分析和SVM-RFE分析筛选出1个核心基因AOX1,通过ROC曲线分析发现AOX1核心基因具有较高的诊断价值,在疾病样本中基因表达比正常样本显著下调.利用Coremine Medical筛选出核桃仁、马齿苋等5味中药,分子对接结果显示原阿片碱、β-胡萝卜素、齐墩果酸、薯蓣皂素匹配度最高.结论 通过生信分析初步预测了与前列腺癌相关的关键基因AOX1,并筛选了具有潜在治疗前列腺癌作用的中药和化合物,以期为临床运用中药治疗前列腺癌提供参考.
Objective To screen the key genes of prostate cancer,and to predict the therapeutic effect of traditional Chinese medicine and compounds.Methods The datasets GSE26910 and GSE55945 were downloaded from TCGA database and GEO database,respectively.Differential expression analysis,weighted gene co-expression network analysis,LASSO regression analysis,and SVM-RFE analysis were performed using R software to screen for core genes in prostate cancer.GO functional enrichment analysis and KEGG pathway enrichment analysis were subsequently conducted.Finally,Coremine Medical and TCMSP were utilized for the screening of traditional Chinese medicines and active compounds,with molecular docking further employed for validation.Results Finally,691 DEGs were identified,which were mainly enriched in biological processes and pathways in muscle contraction,muscle system processes,vascular processes in the circulatory system,mesenchyme development,smooth muscle contraction,and regulation of blood circulation.According to WGCNA analysis,155 key module genes were obtained,and 138 hub genes were obtained by intersection of DEGs and 155 key module genes.Based on LASSO regression analysis and SVM-RFE analysis,AOX1 was selected out.Moreover,ROC curve analysis revealed that the AOX1 core gene had high diagnostic value,and the gene expression in the disease samples was significantly lower than that in the normal samples.Five traditional Chinese medicines such as Walnut kernel and Portulaca oleracea were screened out in Coremine Medical,and the molecular docking results showed that Fumarine,β-carotene,oleanolic acid and diosgenin had the highest matching degree.Conclusion Through bioinformatics analysis,this study preliminarily identifies AOX1 as a key gene associated with prostate cancer and screens a number of traditional Chinese medicines and compounds that may exert potential therapeutic effects against prostate cancer,thereby providing a reference for the clinical application of Chinese herbal medicine in the treatment of this malignancy.
郑瑞蓉;吴姝璐;张伟云;林亚娟;秦飞
厦门医学院药学院/厦门市中药生物工程重点实验室,福建 厦门 361023厦门医学院药学院/厦门市中药生物工程重点实验室,福建 厦门 361023厦门医学院药学院/厦门市中药生物工程重点实验室,福建 厦门 361023厦门医学院药学院/厦门市中药生物工程重点实验室,福建 厦门 361023厦门医学院药学院/厦门市中药生物工程重点实验室,福建 厦门 361023
医药卫生
前列腺癌生物信息学机器学习核心基因中药预测
Prostate cancerBioinformaticsMachine learningCore genesPrediction of traditional Chinese medicine
《医学信息》 2026 (18)
1-8,14,9
1.厦门市自然科学基金项目(编号:3502Z202471059)2.福建省自然科学基金项目(编号:2024J08336)3.厦门医学院大学生创新创业训练计划资助项目(编号:202412631042)
评论