网站首页            期刊简介             编委会             投稿指南             期刊订阅             下载中心             在线留言            联系我们             English
  2025年5月8日 星期四  
文章快速检索
中国生物医学工程学报  2020, Vol. 39 Issue (1): 40-49    DOI: 10.3969/j.issn.0258-8021.2020.01.06
  论著 本期目录 | 过刊浏览 | 高级检索 |
基于Tree-Based LASSO的微生物组子结构回归分析
许小敏, 林勇*
(上海理工大学医疗器械与食品学院,上海 200093)
Regression Analysis of Microbial Substructure Based on Tree-Based LASSO
Xu Xiaomin, Lin Yong*
(School of Medical Instrument and Food Engineering, University of Shanghai for Science and Technology, Shanghai 200093, China)
全文: PDF (12443 KB)   HTML (1 KB) 
输出: BibTeX | EndNote (RIS)      
摘要 人体微生物成分和功能变化对其表型或疾病有着重要的影响,在研究微生物对人体影响时不能仅仅考虑单个微生物动态,还应考虑分类水平下群落的整体影响。为此,提出一种基于tree-based LASSO的微生物组子结构回归分析方法,以分析微生物群落与人体表型之间的关联影响。首先,结合系统发育树结构,构建一种新的惩罚函数逐节点分析树结构;其次,对148个样本进行复杂和稀疏子结构回归对比实验及系数评估,对位于不同子结构上菌种的回归系数进行结果对比分析,并与传统LASSO方法进行比较。结果表明,该方法能够突出微生物群落树结构的影响。在测试节点上的回归系数分别为0.122和0.127,优于传统LASSO方法的回归系数0.106和0.118,从而验证该方法识别菌落结构的优势,因此能更好地分析微生物群落与人体表型之间的关联。
服务
把本文推荐给朋友
加入我的书架
加入引用管理器
E-mail Alert
RSS
作者相关文章
许小敏
林勇
关键词 微生物组LASSO系统发育树关联分析    
Abstract:Human microbial composition and function changes have an important impact on their phenotype or disease. When studying the association of microorganisms with human phenotypes or diseases, not only the individual microbial dynamics, but also the overall impact of the community at the taxonomic level should be considered. In this work, a method of regression analysis of microbial substructure based on tree-based LASSO was proposed to analyze the correlation between microbial community and human phenotype. First, a new penalty function was constructed based on phylogenetic tree structure, and the tree structure is analyzed node by node. Second, 148 samples were tested for complex and sparse substructure regression and coefficient evaluation. The regression results of strains in different substructures were analyzed and compared with the traditional LASSO method. The results showed that this method could highlight the tree structure of microbial communities. The regression coefficients of this method on test nodes were 0.122 and 0.127, which were better than those of the traditional LASSO method (0.106 and 0.118). The advantage of this method in identifying microbial structure was verified. In conclusion, the method could better analyze the association between microbial communities and human phenotypes or diseases.
Key wordsmicrobiome    LASSO    phylogenetic trees    association analysis
收稿日期: 2018-11-15     
PACS:  R318  
基金资助:国家自然科学基金(31301092)
通讯作者: E-mail: yong_lynn@163.com   
引用本文:   
许小敏, 林勇. 基于Tree-Based LASSO的微生物组子结构回归分析[J]. 中国生物医学工程学报, 2020, 39(1): 40-49.
Xu Xiaomin, Lin Yong. Regression Analysis of Microbial Substructure Based on Tree-Based LASSO. Chinese Journal of Biomedical Engineering, 2020, 39(1): 40-49.
链接本文:  
http://cjbme.csbme.org/CN/10.3969/j.issn.0258-8021.2020.01.06     或     http://cjbme.csbme.org/CN/Y2020/V39/I1/40
版权所有 © 2015 《中国生物医学工程学报》编辑部
本系统由北京玛格泰克科技发展有限公司设计开发