Weighted two-band target entropy minimization for the reconstruction of pure component mass spectra: Simulation studies and the application to real system

Abstract

Based on the recently developed algorithm, band-target entropy minimization (BTEM), an algorithm especially suitable for reconstructing discrete pure spectra (such as MS) from mixture systems was developed. It used 1) Peaksa?? Height instead of their derivatives in objective function. 2) Weighting VT vectors to lower the effect of noise on systems. 3) A two-peak targeting method to deal with highly overlapped spectra. 4) An exhaustive searching method to find all components automatically. A global optimization method called fast multi-start simplex method (FMSS) was developed. By discarding some un-promising points in advance, the method dramatically speeded up searching speed in MS reconstructing compared with simulated annealing.These newly developed algorithms were successfully tested on simulated data which contained ten components. On applying them to real systems, with the aid of some experimental strategies, a real MS mixture, which contains four components, with highly noise level was successfully studied.