Dynamic binning peak detection and assessment of various lipidomics liquid chromatography-mass spectrometry pre-processing platforms

Abstract

Liquid chromatography-mass spectrometry (LC-MS) based lipidomics generate a large dataset, which requires high-performance data pre-processing tools for their interpretation such as XCMS, mzMine and Progenesis. These pre-processing tools rely heavily on accurate peak detection, which depends on setting the peak detection mass tolerance (PDMT) properly. The PDMT is usually set with a fixed value in either ppm or Da units. However, this fixed value may result in duplicates or missed peak detection. Therefore, we developed the dynamic binning method for accurate peak detection, which takes into account the peak broadening described by well-known physics laws of ion separation and set dynamically the value of PDMT as a function of m/z. Namely, in our method, the PDMT is proportional to [Formula] for FTICR, to [Formula] for Orbitrap, to m/z for Q-TOF and is a constant for Quadrupole mass analyzer, respectively. The dynamic binning method was implemented in XCMS [1,2] and the adopted source code is available in the Appendix. Our further goal was to compare the performance of different lipidomics pre-processing tools to find differential compounds. We have generated set samples with 43 lipids internal standards differentially spiked to aliquots of one human plasma lipid sample using Orbitrap LC-MS/MS. The performance of the various pipelines using aligned parameter sets was quantified by a quality score system which reflects the ability of a pre-processing pipeline to detect differential peaks spiked at various concentration levels. The quality score indicates that the dynamic binning method improves the performance of XCMS (maximum p-value 9.8{middle dot}10-3 of two-sample Wilcoxon test). The modified XCMS software was further compared with mzMine and Progenesis. The results showed that modified XCMS and Progenesis had a similarly good performance in the aspect of finding differential compounds. In addition, Progenesis shows lower variability as indicated by lower CVs, followed by XCMS and mzMine. The lower variability of Progenesis improve the quantification, however, provide an incorrect quantification abundance order of spiked-in internal standards.