The match distribution quantitation is of use where you have normalised your data as best you can using tools like total sequence count correction, or maybe percentile normalisation quantitation, but you are still seeing slight differences in the shape of the distributions in your data.
It is possible to remove variations in distributions by simply converting your quantitated values to ranks, and then comparing these. The problem with this approach is that this conversion loses any information about the scale of changes which happen in different parts of your distribution.
The match distribution quantitation works by creating an averaged distribution of quantitated values across all of your data sets. It then maps this distribution onto each individual dataset such that they all follow exactly the same profile. This provides much of the same benefits as ranking the data, but preserves the magnitude of the differences at different parts of your distribution.
Where you had repeated values in your original data these will still have identical values in the matched distribution created from the mean of the average distribution positions covered by your range of identical values.
This correction is appropriate to use where you have generally similar distributions, but with some minor variations. If you want to compare distributions which are markedly different you would be better to use the rank quantitation instead.
You can choose a subset of data stores which will be normalised in this analysis.