NMR Peak Alignment Through Lineshape Modeling and Dynamic Programming

• Activity Number: 221
• Type: Contributed
• Date/Time: Monday, August 3, 2009 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Cheng Zheng*+ and Shucha Zhang and Daniel Raftery and Olga Vitek
• Companies: Purdue University and Purdue University and Purdue University and Purdue University
• Keywords: Nuclear Magnetic Resonance (NMR) ; Dynamic Programming ; Peak Alignment ; Voigt profile

Abstract

The 1H NMR spectroscopy is a well-established one dimensional technique widely used in metabolomics. Determination of features that are differentially abundant between disease groups is the primary goal. Since peaks are overlapped and frequency shifted, feature extraction has been a challenge. Binning or clustering solely by peak location does not yield satisfactory results. We model spectra as the summation of components in the shape of Vought profiles first, and then apply dynamic programming to align them using both the peak location and the peak lineshape. Further refinement is achieved by iteration of these two steps and the incorporation of coupling information. The aligned peak list is easy to interpret and opens the possibility of using NMR spectrum database. We show the algorithm performs better than adaptive binning in terms of classification on an experimental NMR database.