"Global median" function in R/Bioconductor?

[shuang]

Which function/package in R/Bioconductor can perform global median?

My input file is a txt separated by tab. My goal is to make/normalize each column to the same median.

[jmw86069]

Try the 'limma' package methods like normalizeBetweenArrays(). If given a numeric matrix, it allows a few methods which aren't array-specific. The method 'scale' will scale to the median of each column. But if for comparison you'd like to try quantile or cyclic loess (depending upon the type of data you're analyzing) they're also available.

Of course, due diligence still applies, in terms of evaluating the method and output of the normalization method, but you're probably watching that closely already. At least some type of boxplot would be good to review, though my preference is something more like a violin plot which shows the shape of the distributions, and can give a clearer view on whether something weird is going on. For violin plots, I use the lattice package, and the panel.violin() extension to the bwplot() function -- but more direct way to test it out is probably to use the vioplot package and its vioplot() function. It would be good to plot before/after normalization to watch exactly what's happening.