ABSTRACT NMR spectroscopy is generally performed in aqueous solutions, thus a frequently occurring problem is the recovering of weak resonances superimposed by an intense solvent signal. Recently, the fully automated approach AUREMOL-SSA/ALS has been successfully applied to different types of n-dimensional NMR spectra; it uses singular spectrum analysis (SSA) for solvent artifact removal combined with an automated linear spline (ALS) for baseline correction. Here, independent component analysis (ICA) is introduced as a new method for the suppression of the solvent signal and compared with SSA. In principle, ICA can overcome some limitations of the SSA but requires a suitable experimental acquisition protocol for its application to one-dimensional NMR spectra. SSA is usually applied to the time domain signal (FID) and requires as input a single FID. In contrast, ICA is optimally applied to frequency domain signals and requires as input at least two different spectra. Here, different acquisition schemes tailored for ICA have been applied to one-dimensional synthetic and experimental datasets of the small globular protein HPr as well as on urine spectra. Excellent results have been obtained by the ICA especially under conditions where SSA fails.
View Full Article
|