Biolog ECO和DGGE数据几种分析方法的比较研究

Comparison among Several Methods for Biolog ECO and DGGE Data Analysis

PCA(Principal Component Analysis)常用于Biolog ECO和DGGE数据分析,但是该方法无法正确区分不同环境微生物的多样性结构,也无法实现微生物标记的发现。为实现该功能,研究采用PCA、PLS-DA(partial least squares-discriiminate analysis)、PLS-EDA(partial least squares-discriiminate enhance analysis)及PLS(partial least squares)、OPLS(orthogonal to partial least squares)方法对Biolog ECO和DGGE数据进行分析。结果表明:DGGE数据通过PLS-EDA分析方法能区分不同环境微生物多样性的结构(PC1=16.8%);采用PLS-DA分析方法,发现两个环境样品中有1个样品重合(PC1=33%);PCA分析方法分离效果最差(PC1=27.1%)。Biolog ECO数据通过PLS-EDA分析方法能区分不同环境微生物多样性的结构(PC1=25.5%);PLS-DA分析方法有4个样品重合(PC1=36.3%);PCA分析方法效果最差(PC1=35.1%)。Biolog ECO和DGGE数据进行PLS和OPLS分析方法筛选后,发现多个潜在的碳源及物种,可作为不同环境条件下微生物标记物。可见,PLS-EDA优于PLS-DA及PCA,是微生物研究的重要方法;PLS和OPLS分析方法中VIP(variable important value)≥1.00的条带和碳源可作为潜在的微生物标记。图7,表1,参24。

PCA is a normal method used for Biolog ECO and DGGE data analysis. However,this method hardly distinguishes soil microbial diversity structure and microbial markers in different environments. In this study,PCA,PLS-DA,PLS-EDA methods were used to study microbial diversity structure in different environments. PLS,OPLS methods were used to discover microbial makers. The results showed that： PLS-EDA method completely separated microbial diversity structure in different environments（ PC1 = 16. 8%）,one sample could not be completely separated by PLS-DA（ PC1 = 33%）,but PCA was not good method to microbial diversity structure for DGGE data analysis. PLS-EDA completely separated microbial diversity structure different environments（ PC1 = 25. 5%）,one sample could not be completely separated by PLS-DA（ PC1 = 36. 3%）,also,PCA was not good method to microbial diversity structure for Biolog ECO data analysis. PLS and OPLS were selected for microbial makers,both methods were suited for underlying microbial makers. In conclusion：PLS-EDA was an optimum method to distinguish different environments soil samples. PLS and OPLS may develop as methods to select underlying microbial makers.