One of the challenges faced in the field of lipidomics is that approximately 75-85% of mass spectral features are unidentified in complex liquid chromatography-tandem mass spectrometry (LC-MS) datasets, colloquially referred to as “dark matter”. Using available LC-MS data processing and characterization techniques, a protocol was developed to minimize the dark matter observed in complex lipidomic datasets and increase the number of lipids identified. Lipids were extracted from bovine liver and analyzed in both polarities via reversed phase HPLC-ESI-MS/MS. Extracted ion chromatograms for each feature in the dataset were extracted using MZmine and annotated using a modified LipidBlast library. Once a preliminary feature annotation was conducted, the data were then further analyzed using a novel protocol employing additional informatics tools including the molecular networking software Global Natural Products Social (GNPS), a Diagnostic Fragmentation Filtering (DFF) tool, a Kendrick Mass Defect (KMD) tool, as well as the LIPID MAPS database to identify additional lipid features. Through this approach, 48 more lipids were added to our list in positive ion mode (335 lipids) and 7 more lipids were added to our list in negative ion mode (214 lipids). This approach also enabled the identification of other spectral features including dimers, polymers, sodium adducts, sodium formate adducts, as well as in source fragmentation peaks. This work demonstrates that the use of multiple data processing and informatics tools reduces the percentage of a dataset that is classified as dark matter by increasing the number of lipid identifications and better characterizing non-lipid features.
Radnoff et al. (Mon,) studied this question.