How is normalisation performed in fractionated LC-MS experiments?

In any LC-MS experiment each run will be affected by systematic errors (e.g. sample loading differences) which will add bias to a quantitative analysis. The purpose of normalisation is to calculate a correction factor which will compensate for these systematic errors.

For fractionated experiments the correction happens in 2 parts. First, we calculate a within-fraction normalisation factor for each run and then a between-fraction normalisation factor for each fraction. Normalised feature abundances for each run are then given as the raw abundances multiplied by the appropriate 2 factors for that run.

The within-fraction normalisation factors are calculated in the first phase of the analysis in which each fraction is analysed separately.

The between-fraction normalisation factors are calculated in the recombine fraction part of the workflow so as to make the mean log normalised abundances equal for each fraction.