生物信息学联合机器学习筛选幽门螺杆菌相关萎缩性胃炎的生物标志物

Bioinformatics combined with machine learning for screening biomarkers in Helicobacter pylori-associated atrophic gastritis

摘要:

目的 利用加权基因共表达网络分析（WGCNA）、机器学习算法筛选幽门螺杆菌相关萎缩性胃炎（HPAG）潜在的生物标志物。方法 下载基因表达数据库中包含HPAG和无幽门螺杆菌感染（nonHP）的胃组织转录组数据进行差异分析，对差异表达基因（DEGs）进行基因集富集分析（GSEA）。整合WGCNA结果和DEGs，筛选HPAG相关基因。利用最小绝对收缩和选择算子（LASSO）、支持向量机递归特征消除（SVM-RFE）和随机森林（RF）等机器学习方法筛选HPAG的潜在生物标志物，提取生物标志物的表达量进行组间比较。结果 共获得213个DEGs，主要富集在胆固醇代谢、脂肪的消化吸收等信号通路。机器学习算法筛选出AF的潜在生物标志物S100钙结合蛋白G（S100G）。HPAG样本中S100G表达水平高于nonHP样本。结论 HPAG发病涉及胆固醇代谢、脂肪的消化吸收等信号通路，S100G在HPAG胃组织中表达显著增高，可能成为HPAG治疗的新靶点。

Abstract:

Objective To screen potential biomarkers of Helicobacter pylori-associated atrophic gastritis (HPAG) using weighted gene co-expression network analysis (WGCNA), and machine learning algorithms.Methods To download the transcriptomic data of gastric tissues containing HPAG and non-Helicobacter pylori (nonHP) infection was from gene expression databases for differential analysis, and perform gene set enrichment analysis (GSEA) on differentially expressed genes (DEGs). WGCNA results and DEGs were integrated to screen HPAG-related genes. Machine learning methods such as least absolute shrinkage and selection operator (LASSO), support vector machine recursive feature elimination (SVM-RFE) and random forest (RF) were utilized to screen potential biomarkers for HPAG, and biomarker expressions were extracted for intergroup comparison.Results A total of 213 DEGs were obtained, which were mainly enriched in signaling pathways such as cholesterol metabolism, digestion and absorption of fat. A machine learning algorithm screened the potential biomarker of AF, S100 calcium-binding protein G (S100G). The expression level of S100G was higher in HPAG samples than in nonHP samples.Conclusion HPAG pathogenesis involves cholesterol metabolism, digestion and absorption of fat, and other signaling pathways. S100G expression was significantly increased in HPAG gastric tissues, which may become a new target for HPAG treatment.