The most popular methods of reducing the influence of shifting peaks are so-called binning or bucketing methods, which reduce the resolution of spectra.
Thereby the spectra are integrated within small spectral regions, which are called “bins” or “buckets”.
Subsequent data analysis procedures, which are applied to the binned spectra, are not influenced by peak shifts,
as long as these shifts remain within the borders of the corresponding bins.
The vast majority of 1H-NMR metabonomics literature uses an equidistant binning of 0.04 ppm.
This means that the spectrum is split into evenly spaced integral regions with a spectral width of 0.04 ppm.
In addition, it is common practice sum up bins covering the citrate doublets (two bins per doublet) into super-bins,
as the shifts of the two citrate doublets span more than one bin width each.
Wide-spread binning with a bin width of 0.04 ppm applied to a set of 1H-NMR spectra of 30 samples from a human metabonomic study. The bin borders are marked by green lines. It is visible, that the taurine peak (triplet marked with a red arrow) shifts between two bins.
The major drawback of equidistant binning is the non–flexibility of the boundaries. If a peak crosses the border between two bins, this peak shift can significantly influence the data analysis. For example, when analysing binned spectra by PCA or PLS, peak shifts result in loadings with opposite signs. If the spectra shown in the figure are binned with the wide-spread equidistant binning method with a bin width of 0.04 ppm, the signal of the triplet of taurine is distributed among two bins. The intensity assigned to each of the two bins highly depends on the exact location of the triplet. Thus, significant variance is added to the two bins, which is not related to the concentrations of taurine and para-hydroxyphenylacetate (singlet with a constant location at 3.455 ppm). Therefore the analysis of the concentrations of these metabolites on the basis of these bins is hampered.