Is it possible to reverse a VST transformation to obtain original counts?

Question

[arashsltni]

I have a VST-transformed matrix and the corresponding size factors from a previous analysis. Is it possible to reverse this back to the original raw count data? My understanding is that the process is likely irreversible. Still, I’m curious if anyone has explored ways to invert or approximate this process.

Answer 1

[ATpoint]

No, the variance stabilization is not a single linear factor you can reverse -- that's the whole point of the method actually. I say my usual thing: Why are raw counts not available to you? That makes the end-to-end analysis intrinsically irreproducible.