This tool effectively normalizes certain samples against some other samples, and removes the effect of the samples normalized to form the effect of the samples that are normalized. This reimplements the same functionality from GeneSpring.
This tools can be used after some other, chip type specific normalization. In other words, normalized data is needed for this tool to work.
One column of the phenodata should be filled with 0's and 1's. The samples used for normalization (their effect is removed) should be coded with 0's, and the samples that are normalized should be coded with 1's. Select this column from the parameters during the normalization.
A separate average expression value for every probe ("identifier") on the samples used for normalization is first calculated. These values are subtracted (since we are working on log-scale) from the values on the to-be-normalized chips.
A new normalized dataset.