transformAssay input

[antagomir]

Extend transformAssay so that it accepts multiple transformations at once?

In addition to
transformAssay(tse, ..., method="clr")

we could have
transformAssay(tse, ..., method=c("relabundance", "clr", "log10"), pseudocount=TRUE)

i.e. method could also be a vector.

Then pseudocount is something we would only like to apply to those transformations where it is really necessary. Here, not to relabundance but yes to clr and log10. One option would be to give that also as a vector, like pseudocount=c(FALSE, TRUE, TRUE).

[antagomir]

@jepasan if you could check this as well when back that would be great

[Daenarys8]

perhaps a wrapper function like transformMultiAssay with method(s) param in for loop could do the trick? or should we rather pass the set in transformAssay?

since we are passing several methods, imo the plural/multi would be intuitive

[antagomir]

Could we just use transformAssay, and allow providing multiple options in methods so they could be all done?

[Daenarys8]

Could we just use transformAssay, and allow providing multiple options in methods so they could be all done?

okay sounds good

[TuomasBorman]

In my opinion, this makes the function too complex without adding much additional benefits. I would rethink implementing this

[antagomir]

Reimplementation can be useful if there is a need.

Adding support for multiple assays pops up on a regular basis and this would be a small change, just for looping over the listed assays?

[antagomir]

Regarding the discussion in #429 we might like to reconsider, for instance, whether altexps need to be supported in the function arguments. It does not simplify the procedures very much compared to doing this outside of the function.