基于生物信息学和机器学习探究非梗阻性无精子症的免疫学机制  

Immunological mechanism of non-obstructive azoospermia:An exploration based on bioinformatics and machine learning

作　　者：黄树强 李志宏 谭翠钰 陈淼琪 袁晓珺 陈婉茹 杨洛瑶 冯许诺 陈彩蓉[1,3] 颜秋霞 HUANG Shu-qiang;LI Zhi-hong;TAN Cui-yu;CHEN Miao-qi;YUAN Xiao-jun;CHEN Wan-ru;YANG Luo-yao;FENG Xu-nuo;CHEN Cai-rong;YAN Qiu-xia(Center for Reproductive Medicine,the Affiliated Qingyuan Hospital(Qingyuan People's Hospital),Guangzhou Medical University,Qingyuan,Guangdong 511518,China;The Third Clinical Medicine School of Guangzhou Medical University,Guangzhou,Guangdong 511436,China;Guangdong Engineering Technology Research Center of Urinary Continence and Reproductive Medicine,Qingyuan,Guangdong 511518,China)

机构地区：[1]广州医科大学附属清远医院/清远市人民医院生殖医学中心,广东清远511518 [2]广州医科大学第三临床学院,广东广州511436 [3]广东省尿控及生殖医学创新工程技术研究中心,广东清远511518

出　　处：《中华男科学杂志》2024年第12期1059-1067,共9页National Journal of Andrology

基　　金：广东省基础与应用基础研究基金(2023A1515220129);广东省中医药局科研项目(20241387);广州医科大学科研能力提升计划项目(2024SRP195);广州医科大学附属清远医院/清远市人民医院开放课题基金(202301-306);广州医科大学2023年度学生创新能力提升计划项目(240603131131)。

摘　　要：目的:基于生物信息学和机器学习探究非梗阻性无精子症(NOA)患者精子发生障碍的免疫学机制,并筛选出精子发生障碍的关键基因。方法:从GEO数据库获得NOA相关数据集,采用差异分析与WGCNA分析得到差异基因,建立精子发生评分模型并进行免疫微环境和细胞通讯分析探究精子发生障碍的机制。通过机器学习筛选关键基因,分析关键基因与T细胞和巨噬细胞的相关性。构建NOA小鼠模型进行转录组测序验证。结果:利用75个差异基因建立精子发生评分模型,精子发生低评分组免疫浸润水平更高,巨噬细胞与T细胞的比例增加,组内细胞互作信号主要与免疫相关。机器学习鉴定SOX30、KCTD19、ASRGL1和DRC7为精子发生的关键基因,在NOA组表达水平下降。关键基因表达水平与T细胞和巨噬细胞浸润水平呈负相关。使用NOA小鼠睾丸转录组测序数据成功验证基于公共数据集构建的精子发生评分模型和机器学习模型的准确性,并成功验证关键基因表达水平趋势。结论:NOA的发生与睾丸免疫微环境增强密切相关,T细胞与巨噬细胞可能在精子发生障碍中起到重要作用,SOX30、KCTD19、ASRGL1和DRC7是NOA诊断和治疗潜在的生物标志物。Objective:To explore the immunological mechanisms underlying spermatogenetic malfunction in patients with non-obstructive azoospermia(NOA)based on bioinformatics and machine learning,and to screen out the key genes associated with spermatogenesis failure.Methods:NOA-related datasets were obtained from the GEO database,and the differentially expressed genes identified by differential analysis and weighted gene co-expression network analysis(WGCNA).A model of spermatogenesis scoring was established for analysis of the immunological microenvironment and cell interaction networks related to spermatogenesis failure.The key genes were screened out by machine learning,followed by analysis of their correlation with T cells and macrophages.An NOA mouse model was constructed for validation of transcriptome sequencing.Results:Seventy-five differentially expressed genes were identified for the establishment of the spermatogenesis scoring model.The low spermatogenesis score group showed a higher infiltration of the immune cells,with an increased proportion of T cells and macrophages and a correlation of cell interaction signals with immunity.SOX30,KCTD19,ASRGL1 and DRC7 were identified by machine learning as the key genes related to spermatogenesis,with down-regulated expressions in the NOA group,and their expression levels negatively correlated with the infiltration of T cells and macrophages.The accuracy of the spermatogenesis scoring and machine learning models,as well as the trend of the expression levels of the key genes,was successfully validated with the transcriptome sequencing data on the NOA mouse testis.Conclusion:The development of NOA is closely associated with enhanced immunological microenvironment in the testis.T cells and macrophages may play important roles in spermatogenesis failure.SOX30,KCTD19,ASRGL1 and DRC7 are potential biomarkers for the diagnosis and treatment of NOA.

关 键 词：非梗阻性无精子症 免疫微环境 生物信息学 机器学习 

分 类 号：R698.2[医药卫生—泌尿科学]