Comparison of MALDI-TOF mass spectrometry data preprocessing techniques and their effect in sample classification

Matrix assisted laser desorption ionization time-of-flight (MALDITOF) is one of the highthroughput mass spectrometry (MS) technologies, producing data requiring an extensive preprocessing before subsequent analyses. In the last years, various preprocessing techniques have been developed for different tasks, including baseline correction, smoothing, normalization, peak detection and peak alignment. Here we present a systematic comparison of various packages for preprocessing MALDI-TOF data. Data is exhaustively preprocessed by using multiple configurations of each preprocessing technique and used to train a set of classifiers, whose performance (kappa and accuracy) results give us the measures for the comparison. We analyze these results in order to study the effect of the preprocessing configurations on classification. The results show that MassSpecWavelet provides the best performance. Sencondary, results also show that SVMs were the most accurate classifiers in our tests.