Lipids are small molecules (mostly <1,500 Da) and are species independent. Composition of a sub-class of lipids might differ from one species to the other but for analytical purposes they are not different. For example, cholesterol is the same lipid regardless of the animal species.

Phospholipids, as a category, are not just one but many classes and subclasses of lipids that consists of phosphatidylcholines, phosphatidylethanolamines, phosphatidylglycerols, etc. Each class of these lipids are unique and require different analytical strategies. Hence, they cannot be analyzed together. One has to specify which class and subclass of lipids to be analyzed before cost can be estimated.

This is the most difficult question to answer without empirical data. All our lipidomic analyses utilize Liquid Chromatography – Mass Spectrometry (LC-MS). Sensitivities of LC-MS methods are described as ‘quantity (ng, pg, etc.) on the column’. For example, if a lipid has a detection limit of ‘10 pg on the column’, that much lipid must be present in the sample processed for detection. Since lipids have to be extracted from the sample before analysis, the detectable quantity of the lipid can be present in any volume of sample collected (typically 10 µl to 3 ml of fluids and homogenates containing 0.1-2 mg protein; too large a sample volume contributes to higher background noise). Thus, low levels of lipids require larger volumes and vice versa. Also, other abundant lipids present in the sample that are extracted with the lipid of interest can potentially mask the low abundance lipid detection.

In each sub-class of lipids (e.g. fatty acids, phosphatidylcholines, etc.), there are always a few molecules that are at much higher concentrations than the rest (as much as 10⁵). For example, palmitic acid can be 10,000 times more than docosahexaenoic acid in a cell homogenate. Despite the large dynamic range of mass detectors (10³-10⁵), when the sample is diluted to bring the most abundant lipid into the dynamic range of the detector for quantification, the low abundance lipids are too diluted to be detectable. As a result, the sample has to be analyzed twice, once at higher concentration to quantify the low abundance lipids and a second time at a lower concentration to quantify the more abundant lipids. Obviously, this adds to the cost of analysis when the entire range of molecules in a sub-class of lipids have to be accurately quantified.

Lipidome (metabolome in general) is significantly more prone to variation compared to the proteome and genome. Even small changes in experimental conditions such as temperature, time, pH, buffer components, etc., can cause large changes in select compounds of the lipidome. As a result, experimental conditions between replicates should be matched as closely as possible. Despite all precautions, variations between samples can be high. In addition, technical variation inherent to LC-MS methods can contribute 5-15% CV. Where internal standards are used for LC-MS (most of our methods are internal standards based), the CV is typically 3-5%. Following are the general guidelines for sample sizes:

 

Sample type	n (range)
Tissue cultures	3-5
Controlled animal experiments	9-15
Clinical samples	15-25+

 

   The suggested sample size is not mandatory. Users are required to assess their own requirements in consultation with an expert statistician.

Unless specifically required by the user (and paid for!), all our quantitations are relative to the internal standards used. We use a set of internal standards that match the chemical structure and HPLC retention times as closely as possible to the analytes of interest. This method assures that the extraction of analytes from sample closely matches the standard against which it is quantified. Because we use the same internal standards for the same analytes across all samples (as well as across different batches of samples), the relative quantitation is highly accurate to compare between analyses. However, we strongly recommend that all samples in an experiment (i.e. both control and experimental samples) be analyzed together for best statistical outcome.