中文摘要：

基于核磁共振技术的代谢组学是近年发展起来的一种新的组学技术，主要利用生物体液的核磁共振谱提供生物体内全部小分子代谢物的丰富信息。然而，噪声的存在影响了模式识别方法分析的准确度。近年来小波变换以其多分辨率分析的特性、方法简单、快速等优点成为一种有效的去除分析信号噪声的方法。本实验通过运用小波变换去除噪声、校正基线后，再进行Fisher判别分析，得到了较传统分析更为清晰的代谢标识物，建立了良好的代谢模型。

英文摘要：

^1H nuclear magnetic resonance （^1H NMR）based metabonomics and pattern recognition analysis （PRA） as an important component of global systems biology is a new emerging ＂omic＂ technology in recent years. The ^1H NMR spectra provide a wealth of metabolites information in biofluid. However, spectroscopic data of samples are confused by a series of noise, which greatly influences the achievement of accurate analytical result. In this study, wavelet was used to reduce the noise and remove the baseline wonder of IH NMR spectra. Then the sample spectral data were analyzed by the Fisher distinguish analysis which can produced good elassification and obtain remarkable biochemical NMR markers. Compared with traditional methods such as PCA, PLS, this combination of Fisher distinguish analysis and wavelet can be applied to establish a better biologic metabolize models.