I have used both to transform my leukaemia RNA-Seq data for subsequent hierarchical clustering. The result is quite different. Some subtypes of leukaemia only form a cluster (or at least sit closer) in one of the transformed data set. While I understand both can be used to transform counts into gene expression matrix that can be use for clustering, I want to know more about the difference. From my really poor statistic background I assume voom() is very similar to rlog(CPM) and can generate a microarray-like data set. On the other hand, the name of vst() is quite self-explanatory and is described to be better at handling varying size factors than rlog() does.
What are the actual difference and is there one that you think is better at handling jobs like clustering analysis?
I am hoping this to be a more educational question than technical one. If it violates any rules please delete it.