Blind Separation of Analytes in Nuclear Magnetic Resonance Spectroscopy and Mass Spectrometry: Sparseness-Based Robust Multicomponent Analysis

摘要

Metabolic profiling of biological samples involves nuclearnmagnetic resonance (NMR) spectroscopy and mass spectrometryncoupled with powerful statistical tools for complexndata analysis. Here, we report a robust, sparsenessbasednmethod for the blind separation of analytes fromnmixtures recorded in spectroscopic and spectrometricnmeasurements. The advantage of the proposed methodnin comparison to alternative blind decomposition schemesnis that it is capable of estimating the number of analytes,ntheir concentrations, and the analytes themselves fromnavailable mixtures only. The number of analytes can benless than, equal to, or greater than the number ofnmixtures. The method is exemplified on blind extractionnof four analytes from three mixtures in 2D NMR spectroscopynand five analytes from two mixtures in massnspectrometry. The proposed methodology is of widespreadnsignificance for natural products research and the fieldnof metabolic studies, whereupon mixtures representnsamples isolated from biological fluids or tissue extracts.