Advise how VST will transform data
vst_advice(what = "anscombe.nb", dispersion = NULL, cpm = FALSE, lib.size = NULL)
what | Either the output of a call to vst() or the name of a VST method (see vst() help). |
---|---|
dispersion | As per vst(). |
cpm | As per vst(). |
lib.size | As per vst(). |
A data frame giving an indication of how an average sample will be transformed.
The column "twofold_step" shows the step from the previous to current row. With a log2 transformation this would be uniformly 1, but with a VST and small counts the step is less than 1. This therefore provides advice on how compacted the VST is near zero-count, as compared to a log2 transformation.
Note that the results given are for an average sample. Where library sizes differ wildly, the VST may perform poorly.