A Comparative Study Of Several NMR Peak Alignment Methods

Peak alignment is an important step during metabolomics data pretreatment process based on nuclear magnetic resonance(NMR), and its effect plays a direct role on subsequent multivariate statistical analysis. NMR spectra peaks have shapes such as singlet, doublet, triplet, multiplet peaks, the common method for doing in-tercepted spectral shift correction is based on the highest point of the calibration peak (max-peak-alignment) methods. Similarity measure such as distance mea-sure and similarity measure are widely used in image registration, in this paper, we applied these similarity metrics and newly proposed percentage method on the NMR spectra of the various peaks for doing translational correction, and conducted quantitative and comparative analysis over several correction method based on the obtained results. Quantitative analysis showed that:cosine corr guidelines are con-sistent with the criteria correction effect, this is consistent with the theory, because in the case of data normalization, the two are equivalent; L1norm and L2norm are not very different to some extent, But compared to the L1norm, the correction effect by L2norm is better; Relative to the L1, L2, or cosine criterion, percentage correc-tion methods in this article do not show certain advantages, the compared data in the study indicate that, the method has some effect on the percentage of triplet of correction, and max methods in each of the selected peak shape in this article shows the poor performance, according to the definition of the criteria it shows that, max method is related to the shape of the peaks especially the higher symmetry require-ment. This comparative study can do it for reference of the research in correction of spectral peak selection and subsequent multivariate statistical analysis.