A mass spectra-based compound-identification approach with a reduced reference library

Abstract

In this paper, an effective and efficient compound identification approach is proposed based on the frequency feature of mass spectrum. A nonzero feature-retention strategy, and a correlation based-reference library reduction strategy, are designed in the proposed algorithm to reduce the computation burden. Further, a frequency feature based-composite similarity measure is adopted to decide the chemical abstracts service (CAS) registry numbers of mass spectral samples. Experimental results demonstrate the feasibility and efficiency of the proposed method. © 2013 Springer-Verlag.