摘要
痤疮作为常见皮肤病之一,发病机制复杂,其中微生物定植在痤疮发病中的作用是一个热点研究问题。本文从宏基因组学的角度,利用机器学习方法分析痤疮宏基因组数据,包括痤疮患者的患病皮肤(diseased skin,DS)样本集和健康皮肤(healthy skin,HS)样本集,以及正常对照组(normal control,NC)样本集。为了同时分析3组样本集以获得可以区分不同样本集的脂质,使用多重集典型相关分析(multi-set canonical correlation analysis,MCCA)方法进行研究。实验结果可得到仅对某一样本集有显著影响的脂质,以及同时对3个样本集影响程度不同的脂质,这些脂质可以作为判别皮肤状态的指标,用于辅助指导皮肤痤疮疾病的诊断、预后和治疗。
As one of the common skin diseases,the pathogenesis of acne is very complicated.The role of microbial colonization in the pathogenesis of acne is an active research area.The purpose of this paper is to analyze acne metagenomic data,including sample sets of acne diseased skin(DS)and healthy skin(HS)as well as normal control(NC)by using machine learning from the perspective of macrogenomics.Multi-set canonical correlation analysis(MCCA)method is used to analyze the above three sample sets at the same time and to confirm the lipids that can distinguish these three sample sets.The results show that lipids that had a significant impact on only one set and those that had different impacts on the three sample sets respectively can be used as indicators to determine the skin status.Moreover,these lipids can be used to guide diagnosis,prognosis,and treatment of skin acne diseases.
作者
孙梦茹
王瑜
何聪芬
贾焱
高学义
SUN Mengru;WANG Yu;HE Congfen;JIA Yan;GAO Xueyi(Beijing Key Laboratory of Big Data Technology for Food Safety,School of Computer and Information Engineering,Beijing Technology and Business University,Beijing 100048,China;Key Laboratory of Cosmetic of China National Light Industry,School of Science,Beijing Technology and Business University,Beijing 100048,China)
出处
《智能系统学报》
CSCD
北大核心
2020年第5期972-977,共6页
CAAI Transactions on Intelligent Systems
基金
国家自然科学基金面上项目(61671028)
北京市自然科学基金面上项目(4162018)。
关键词
痤疮
宏基因组学
面部皮肤
脂质
机器学习
多重集典型相关分析
acne
macrogenomics
facial skin
lipids
machine learning
multi-set canonical correlation analysis
