Feature Grouping Technique Based on Biclustering for the Analysis of LC-MS Metabolomic Data

	Feature Grouping Technique Based on Biclustering for the Analysis of LC-MS Metabolomic Data
	Lin XH(林晓惠) ; Ruan Q(阮强) ; Zhou LN(周丽娜) ; Yin PY(尹沛源) ; Xu GW(许国旺)
	2011
会议名称	37th international symposium on high performance liquid phase separations and related techniques
会议日期	2011-10-8
会议地点	大连
页码	492-0
通讯作者	许国旺
中文摘要	metabolomics has shown a promising application in many fields such as disease diagnosis, drug development. liquid chromatography-mass spectrometry (lc-ms) is one of its main analysis techniques. hplc-ms metabolomic data is usually of high dimension. among the large features, some are related to each other and contain the similar information about the problem. grouping the features correctly is very meaningful for getting a comprehension of the problem studied and building a more efficient classification model. in this work, a lc-ms dataset which contains serum specimens from 30 normal samples, 30 hepatitis patients (h), 30 cirrhosis patients (c) and 30 liver cancers patients (t) was got. to distinguish the different kinds of the liver disease, we proposed an ensemble classification method based on the feature grouping by the biclustering [1] technique (ec-bicfg). for each base classifier, the feature subspace is generated according to the group ranking. naive bayes (nb) and 5-nearest-neighbor (5nn) are adopted as the base classifiers, respectively. in addition to discriminating between controls and patients, we also conducted the experiments to distinguish among three liver diseases, and between each two kinds of the liver diseases. the corresponding accuracy rates are listed in table 1. it shows that our method out performs egsg [2] which is also an ensemble classification algorithm based on feature grouping. table 1 the loocv classification accuracy rates nb (%) 5nn (%) egsg ec-bicfg egsg ec-bicfg h vs. c 80.50(2.61) 87.67(1.61) 71.99(5.49) 84.50(1.77) h vs. t 76.50(5.41) 86.33(2.58) 60.00(5.21) 77.67(1.41) c vs. t 76.50(4.93) 81.00(2.63) 54.00(3.78) 83.83(1.24) h vs. c vs. t 62.33(5.84) 73.11(0.47) 54.33(3.41) 72.11(1.77) control vs. model 95.42(1.85) 96.83(1.61) 89.75(2.69) 97.42(1.07) reference [1] yizong cheng, george m. church. biclustering of expression data. in proceedings of 8th international conference on intelligent system for molecular biology (ismb) (2000) 93–103. [2] huawen liu, lei liu, huijie zhang. ensemble gene selection by grouping for microarray data classification. journal of biomedical informatics 43 (2010) 81–87.
合作状况	墙报
会议主办者	中国化学会色谱专业委员会
会议录	proceeding of hplc 2011
会议录出版者	待补充
会议录出版地	待补充
学科主题	分析化学
内容类型	会议论文
源URL	[http://159.226.238.44/handle/321008/116075]
专题	大连化学物理研究所_中国科学院大连化学物理研究所
推荐引用方式 GB/T 7714	Lin XH,Ruan Q,Zhou LN,et al. Feature Grouping Technique Based on Biclustering for the Analysis of LC-MS Metabolomic Data[C]. 见:37th international symposium on high performance liquid phase separations and related techniques. 大连. 2011-10-8.