Quality control requirements for the correct annotation of lipidomics data
Harald Köfeler, Thomas O. Eichmann, Robert Ahrends, John A. Bowden, Niklas Danne-Rasche, Edward A. Dennis, Maria Fedorova, William J. Griffiths, Xianlin Han, Jürgen Hartler, Michal Holčapek, Robert Jirásko, Jeremy P. Koelmel, Christer S. Ejsing, Gerhard Liebisch, Zhixu Ni, Valerie B. O’Donnell, Oswald Quehenberger, Dominik Schwudke, Andrej Shevchenko, Michael J.O. Wakelam, Markus R. Wenk, Denise Wolrab, Kim Ekroos
Abstract
A recent publication from Vasilopoulou et al. While this represents an impressive technological advance with the potential to increase lipidome coverage and lower detection limits for individual lipids, the interpretation of the acquired spectra is a matter of concern. Specifically, the authors relied exclusively on software-assisted lipid assignments that were not confirmed by an independent inspection of matched spectra to recognize abundant structurally unique lipid fragments. Further, no attempts were made to correlate the retention times of identified species with available lipid standards, which constitutes the gold standard typically employed in lipidomics to reduce false-positive assignments. Manual inspection of the dataset performed by us suggested that the identification of at least 510 out of 1108 features reported as unique lipids would require additional experimental evidence. This, in turn, compromises the assignment of collision cross section (CCS) values for 1856 features, potentially misguiding other lipidomics laboratories that may use these CCS data for identifying lipids.