Peak Detection with Chemical Noise Removal Using Short-Time FFT for a Kind of MALDI Data

Abstract

Peak detection is the first step in biomarker extraction from the mass spectrometry data, which significantly influences the results of the following steps. Designing a good method for peak detection greatly depends on the properties of the data. In this paper, we propose a novel automatic peak detection method without the a priori knowledge on the mass of the proteins for a kind of MALDI data, which have a regular noise pattern called chemical noise except the random noise. The random noise is removed by using the undecimated wavelet transform. An adaptive short time discrete Fourier transform is proposed to do the chemical noise de-noising. We combine the possible peaks corresponding to one protein by extracting an envelope over them. Depending on the signal-to-noise ratio, the desired peaks that have the highest intensity among their peak clusters in each individual spectrum are detected. We examine the performance of the method in the carotid artery disease data set that shows the efficiency of the method. With the chemical noise removal, the signal-to-noise ratio of the peaks is increased greatly compared to the result without chemical noise removal.