Background information about the Shotgun Lipidomics platform. 

Lipids are extracted from biological samples using a modified Bligh-Dyer method [1] using liquid-liquid extraction at room temperature after spiking with internal standards. Analysis of lipids is performed on reversed phase HPLC, followed by MS analysis that alternates between MS and data-dependent MS2 scans using dynamic exclusion in both positive and negative polarity and yields excellent separation of all classes of lipids. Quality Controls (QC) are prepared by pooling equal volumes of each sample, in addition to a well characterized plasma pools, and are injected at the beginning and end of each analysis and after every 10 sample injections, to provide a measurement of the system's stability and performance as well as reproducibility of the sample preparation method [2]. Lipids are identified using the LipidBlast [3] library (computer-generated tandem mass spectral library of 212,516 spectra covering 119,200 compounds from 26 lipid compound classes), by matching the product ions MS/MS data. The method allows us to measure >70 percent of lipids with an intensity RSD value below 20 percent belonging to eight different lipid classes which includes phospholipids like lysophosphatidylcholine (lysoPC), phosphatidylcholine (PC), phosphatidylethanolamine (PE), phosphatidylserine (PS), phosphatidylglycerol (PG), phosphatidylinositol (PI), glycerolipids like triglyceride (TG), diglyceride (DG) and monoglyceride (MG) and sphingolipids like sphingomyelin (SM) and ceramides (Ce) in a single RP UPLC-QTOF MS/MS acquisition. The lipids are quantified using MultiQuant and normalized by internal standards. This method displays excellent reproducibility, mass accuracy and no significant carryover.

Data Analysis

Internal standard are included for each lipid class measured for normalization and quality control purposed. CSVs are generally 10-30 percent.

Lipids with more than 50 percent missing values for tissues and more than 20 percent for cell lines will be removed from further analysis, as well as those with a coefficient of variation (CV) exceeding 5 percent; the remaining missing values will be imputed either at the minimum detection level or through a k=5 nearest neighbor procedure. Primary analysis will detect differentially expressed compounds across experimental groups using parametric (two sided t-test) and non-parametric (Wilcoxon rank-sum) tests for two groups, or analysis of variance (ANOVA) for more than two groups. Family-wise error rate (FWER) methods and false discovery rate (FDR) methods would be employed for multiple hypothesis testing correction.  Supervised learning can be further employed to obtain parsimonious models of association with clinical outcomes (e.g. disease/normal status, etc); for small sample number k-nearest neighbor classifiers and linear discriminate analysis will be employed, while larger sample groups will be analyzed using complex methods such as support vector machines, bagging, boosting, neural networks, and random forests. Multiple classification models will be assesses using K-fold cross-validated error rates, analysis receiver-operator characteristic (ROC) and comparisons of the area under the curve (AUC). Data visualizations will employ principal components analysis (PCA) and hierarchical clustering of samples and/or metabolites.

The method is successfully utilized for the comprehensive lipidomic profiling of complex biological samples like tissues, plasma, serum, urine, saliva, cells, etc.

References

1. Bligh, E. G.; Dyer, W. J. A Rapid Method of Total Lipid Extraction and Purification. Can. J. Biochem. Psysiol 1959, 37, 911–917.

2. Gika, H. G.; Macpherson, E.; Theodoridis, G. A.; Wilson, I. D. Evaluation of the repeatability of ultra-performance liquid chromatography−TOF-MS for global metabolic profiling of human urine samples. J. Chromatogr., B: Anal. Technol. Biomed. Life Sci. 2008, 871 (2), 299−305.

3. Kind, T.; Liu, K.-H.; Lee, D. Y.; DeFelice, B.; Meissen, J. K.; Fiehn, O. LipidBlast in silico tandem mass spectrometry database for lipid identification, Nat. Meth 2013, 10, 755−758.