Lipid+Analysis+Data+Handling

=LIPID ANALYSIS: DEALING WITH DATA, brief overview= University of Wisconsin Madison**
 * T. Balser, Associate Professor Soil and Ecosystem Ecology

Once analyzed by the GC, lipid results must be converted from raw data (peak area) to a form that can be statistically analyzed and interpreted. Unfortunately, this data analysis and reduction can be tedious and time-consuming. We have tried here to provide a step-by-step ‘road map’, coupled with an Excel spreadsheet that demonstrates the calculations. Generally, the main steps are:
 * Overview**:
 * 1) Convert from area to μg C using conversion factor from internal standards run on the GC (conversion factor = μg C of the standard divided by the area of the standard on the GC, and is typically the average of two internal standard lipids, 9:0 and 19:0, to capture the response of both short- and long-chain lipids),
 * 2) Divide by unit weight soil or organic matter, units = μg C g soil-1,
 * 3) Convert to μmol C g soil-1 using the molecular weight, and
 * 4) Use selection criteria to reduce size of the dataset (e.g., reject lipid if it is <0.5% of the dataset; this cutoff tends to preserve relatively rare as well as common lipids but removes extraneous data).

Processed lipid data are expressed as abundance (nmollipid/gsoil), mole fraction (nmollipidx/nmoltotallipid, 0-1), or mole percent (mole fraction*100, 0-100%). The choice of expression depends on the question of interest, but usually examination of the dataset benefits from using all three. Mole fraction and mole percent are normalized by the total biomass in a sample, and are thus measures of the relative abundance of any given lipid. Mole fraction is appropriate for use in ordination analyses (after transformation, e.g. arcsince square-root), while mol% (mole fraction multiplied by 100) is an easy-to-interpret value. Lipid abundances are the absolute amount of a given lipid extracted per gram of soil. Because the quantity of lipid per cell is reasonably constant, and the lipid extraction is highly quantitative (i.e. close to 100% extraction efficiency) abundance is in effect an estimate of microbial biomass. Total abundance is total biomass, and the abundance of key indicators reflects the biomass of the group it represents.


 * Calculations** (see Excel workbook for ‘real’ examples of these):

1) Calculate K, conversion factor
 * Must know the concentration of the internal standard (µg lipid/µl). This is predetermined by the stock and working solutions.
 * Must know the average peak area for internal standard lipids (usually on the order of 6000 area units). This is determined from running blank samples with only the working internal standard (IS) solution as part of each lipid run on the GC. We take an average peak area (response) for all IS samples we run.

Conversion factor, K (µg lipid/µl std)/area x 2µl GC ‘sip’ µg lipid/area

2) Convert from peak area to µg lipid
 * Must know K
 * Must have peak area for each lipid in a given sample (peak area Xi)
 * Must know volume of working solution to which each sample is resuspended (usually approx 250µl)
 * Must know the mass of soil extracted (g soil )

µg lipid /g soil = (peak area Xi * K * µl Resuspended_vol/2µl GC sip)/g soil extracted

3) Convert µg lipid to µmol lipid/g soil
 * Must know molecular weight of each lipid in the sample

µmol lipid/g soil = (µg lipid /g soil)/molwt of lipid

nmol is sometimes a nicer unit for reporting biomass… nmol/g soil = µmol lipid/g soil *1000

4) Calculate mol fraction/mol percent
 * Must know sum of total lipid extracted (usually this is after we go through the data and delete all lipids that look like background noise – only present in one sample, or in very low abundance in only a few samples).

Mol fraction of lipid in sample, lipidi = (µmol lipid/g soil )/total lipid extracted in that sample Mol % is the mol fraction multiplied by 100.

5) Arcsin transform the mol fraction for ordination
 * Must know mol fraction of each lipid in each sample
 * Because mol fraction data are not normally distributed (range is constrained from 0-1), we must arcsin transform the data

Arcsin_lipidi-arcsin(mol fraction^0.5) [negative arcsin of the squareroot of the mol fraction]

6) Use the processed data! I usually sum up lipids into groups, or look at the abundance and relative abundance of specific indicator lipids, and make bar plots as well as plotting the results from ordination. I am working on a description of the ways to use the data – look for it to be posted soon.