物理化学学报 >> 2006, Vol. 22 >> Issue (09): 1052-1055.doi: 10.3866/PKU.WHXB20060903

研究论文 上一篇    下一篇

一种新多肽表征方法及支持向量机用于肽HPLC定量结构-保留建模预测

梁桂兆;李志良;周原;何留;周鹏   

  1. 重庆大学化学化工学院; 重庆大学生物力学与组织工程教育部重点实验室, 重庆 400030
  • 收稿日期:2006-01-10 修回日期:2006-03-21 发布日期:2006-09-04
  • 通讯作者: 李志良 E-mail:zlli2662@163.com

A New Peptide Sequences Representation Technique and Support Vector Machine for Quantitative Structure-Retention Modeling of Peptides in HPLC

LIANG Gui-Zhao;LI Zhi-Liang;ZHOU Yuan;He Liu;ZHOU Peng   

  1. College of Chemistry and Chemical Engineering, Chongqing University, Chongqing 400030, P. R. China; Key Laboratory of Biomechanics and Tissue Engineering, MOE, Chongqing University, Chongqing 400030, P, R. China
  • Received:2006-01-10 Revised:2006-03-21 Published:2006-09-04
  • Contact: LI Zhi-Liang E-mail:zlli2662@163.com

PDF

4872

赞

摘要：

从20种天然氨基酸的1369种性质参数经主成分分析得出一种新多肽序列表征方法——SZOTT. 将其用于71个不同长度肽序列表征, 以偏最小二乘(PLS)和支持向量机(SVM)建立定量结构-保留模型(QSRM). 研究表明, SZOTT能够较好表征71个肽序列特征, 其含信息量大且易操作, 与PLS相比, SVM对lgk建模预测表现出较强的拟合能力和良好外部预测能力, SZOTT表征方法和SVM建模可进一步用于肽HPLC保留行为研究.

关键词: 肽, SVM, QSRM, SZOTT

Abstract:

A new representation technique for peptide sequences, namely SZOTT(scores vector of zero dimension, one dimension, two dimension, and three dimension), was derived from 1369 parameters of 20 coded amino acids using principle components analysis (PCA). It was then employed to express 71 peptide sequences with different lengths. Quantitative structure-retention modelings (QSRMs) were constructed by support vector machine (SVM) and partial least square (PLS). The results indicated that 71 peptide sequences could be preferably represented by SZOTT with many advantages, such as plentiful structural information and easy manipulation. Also simulative power for interior samples and predictive power for exterior samples by SVM were superior to those from PLS. SZOTT and SVM can be applied to develop QSRMs.

Key words: Peptides, SVM, QSRM, SZOTT