The autocorrelation matrix probing biochemical relationships after metabolic fingerprinting with CE.

Santiago Angulo, Isabel García-Pérez, Cristina Legido-Quigley, Coral Barbas

Index: Electrophoresis 30(7) , 1221-7, (2009)

Full Text: HTML

Abstract

Fingerprinting together with statistical analysis is often employed to compare samples in metabonomic studies of a disease. Correlation algorithms can aid by extracting information based on the variation patterns of key metabolites. This information can be linked to metabolite identification or to specific up/down-regulated biochemical pathways. Matlab-based software employing the Pearson's correlation algorithm was applied to urine electropherograms from 20 mice infected with the schistosoma parasite. The fingerprints were the sum of electropherograms analysed with normal and reverse polarity, in two different modes MEKC and CZE and with two different capillaries (uncoated and polyacrylamide coated) to provide a broad picture of the samples. Hippurate, a metabolite that was depleted in the infected group and is present in both polarities, was chosen as a test variable; it correlated with itself to a p value of <0.000. Phenylacetylglycine, a metabolite shown as over expressed in the disease, was positively correlated to three metabolites in its same pathway with a correlation coefficient of 0.7 and p<0.000 to phenylalanine, 0.7 and p<0.000 to 2-hydroxyphenylacetic and 0.55 and p<0.003 to phenylacetate. The study shows that the autocorrelation matrix is able to provide extra information from data files acquired by CE analyses. It underlined an up-regulated metabolic path by association in the schistosoma infection model.