Study on chromatographic data pre-processing using fuzzy decision making in metabonomics

This paper introduced a straightforward and effective chromatographic data pre-processing method developing for utilization prior to chemometric analysis of large metabonomic dataset arising from high performance liquid chromatography. Nucleotides chromatographic fingerprinting in human urines was employed to validate the proposed method. Performance for discrimination cancer samples from healthy urinary sample with principal component analysis was advanced by this method. The first and second principal components could discriminate the two groups by a straight line with precision rate of 75% for cancer samples and 92% for normal samples. Using unprocessed data only 50% cancer samples could be discriminated from healthy cluster. This method was proved to be an effective chromatographic data pre-processing procedure in metabonomics