Date of this Version
Chemometr Intell Lab Syst. 2014 February 15; 131: 1–6. doi:10.1016/j.chemolab.2013.11.005.
Nuclear magnetic resonance (NMR) spectroscopy has proven invaluable in the diverse field of chemometrics due to its ability to deliver information-rich spectral datasets of complex mixtures for analysis by techniques such as principal component analysis (PCA). However, NMR datasets present a unique challenge during preprocessing due to differences in phase offsets between individual spectra, thus complicating the correction of random dilution factors that may also occur. We show that simultaneously correcting phase and dilution errors in NMR datasets representative of metabolomics data yields improved cluster quality in PCA scores space, even with significant initial phase errors in the data.