Proteomic Mass Spectrometry II: Cleaning and Preprocessing

Keith Baggerly
Department of Biostatistics & Applied Mathematics
M.D. Anderson Cancer Center

Abstract:

In order to make sense of mass spectrometry proteomic data,
it is often necessary to "clean" the data so that the final 
contrasts between disease and control spectra will make sense. 
In this talk, we first present a case study involving the study
of MALDI mass spectra derived from serum for the investigation 
of lung cancer, using it to illustrate the need for baseline
subtraction, normalization, and other processing steps. We then
introduce some of the methods that we are currently using to 
process like spectra, ranging from the undecimated wavelet
transform for smoothing and baseline subtraction to the 
use of simple averaging to improve peak detection.