Computational methods for the differential profiling of triacylglycerols using RP-HPLC/APCI-MS

Reversed phase liquid chromatography with atmospheric pressure chemical ionization mass spectrometry (RP-HPLC/APCI-MS) was employed for the analysis of natural mixtures of triacylglycerols. An integrated framework for data analysis, including preprocessing, statistical analysis and automated structure identification, was implemented in the R statistical program. Raw data stored as mzXML, mzData, or mzML files are preprocessed using a series of steps for peak detection, chromatographic alignment, and normalization. Targeted and non-targeted feature selection steps are employed to filter the data for features that are relevant and informative for a particular biological question. Triacylglycerol structures are identified by evaluating relationships between the diacylglycerol fragment ions and protonated molecules observed in APCI mass spectra, and suggested structures are evaluated using a correlation-based score that reflects whether structure-associated ions are concurrently eluting over the retention-time course of the analysis. The algorithm was tested using five soybean oils and triacylglycerol structure identifications were verified from literature references. We employed the developed methodology for classification of plant oils and marine oils to their biological source, and also to determine structural differences in triacylglycerols in adipose tissue from mice fed different high-fat diets in studies of diet-induced obesity.