Metabolomics Data Analysis Improvement by Use of the Filter Diagonalization Method

Abstract

The filter diagonalization method (FDM) was implemented and used instead of fast Fourier transform (FFT) to obtain the nuclear magnetic resonance (NMR) spectra from the free induction decay (FID) signals. The areas obtained by the FDM, from selected absorption lines, were used as input for a multidimensional method of data analysis. This procedure was applied in a NMR-based metabolomics investigation. In FDM, instead of spectra, the absorption peaks’ specification, such as central frequency, line width, amplitude and relative phases, are estimated and the spectra are built using this information. Therefore, one can select the lines by width and intensity to exclude the broad lines such as baseline, solvent line and albumin peak. Also lines with small amplitude such as noise can be excluded from the spectra. Moreover, the spectra do not suffer from aliasing or baseline problems. These characteristics are fundamental in the metabolomics investigations. To show the superiority of our method over the standard FFT to obtain the spectra, we reconstructed the spectra from simulated FID by both methods. As an example, this new approach is used to analyze the non-small cell lung cancer A549 exposed to different treatments and principal component analysis is used to compare the performance of both methods.