Bioinformatics for Quantitative, Targeted Lipidomics Discovery

Abstract

Lipidomics uses high performance liquid chromatography coupled to tandem mass spectrometry via electrospray ionization to comprehensively identify and quantify all lipid species in a biological system. Divided into untargeted and targeted approaches, untargeted lipidomics is viewed as a discovery-driven method to possibly identify all lipids, while targeted lipidomics is considered an approach limited to quantifying known lipids. While there are many bioinformatics tools to facilitate lipid identification in untargeted lipidomics datasets, there are few tools to annotate lipid identities in targeted datasets. Thus, I present two novel bioinformatics tools to annotate and identify lipids from targeted lipidomics data. The first tool annotates lipid isotopes and in-source artifacts caused by electrospray ionization from targeted lipidomics. Using this tool, I highlight problems associated with the manual curation approach conventionally used to pick peaks from targeted lipidomics data. In Bayesian Annotations for Targeted Lipidomics (BATL), I present the first principled and accurate classifier for the identification of targeted lipidomics data. This program eliminates human subjectivity associated with peak identification. Combining these bioinformatics tools together, I conclude by presenting a fully unbiased, discovery-driven, targeted lipidomics pipeline, enabling the identification of all lipid isomers detected from targeted lipidomics data.