Closed lmsimp closed 4 years ago
@lgatto Did we decide that is is okay to call mrkVecToMat
inside new vis apps if input is not a matrix?
Yes, doing that transparently is fine, as long as the user's data is not modified.
One issue, all the feature data columns in the new hyperlopit data, have names that begin with capitals (see data(hyperLOPIT2015)
and fvarLabels(hyperLOPIT2015)
in pRolocdata
). This is very annoying for differentiating between vectors and matrices in the fData
. There is the situation where I have a column called Markers
which is a matrix, but I wish to plot the column SVM
using my app, now, if I call mrkMatToVec
with vfcol = "SVM"
then I get an error as there is already Markers
, I can specify a name for mfcol
e.g. mfcol = "xxx"
and plot that, this is all invisible and not seen my user, what would you suggest I call mfcol
for these cases?
Yes, doing that transparently is fine, as long as the user's data is not modified.
When I call mrkMatToVec
as discussed above, the users data is not altered as we do not return object
. To be extra safe I can create a new MSnSet
i.e. a copy of object
but with the new Markers
matrix appended, and use that for the interactive plotting, if in the future you envisage returning the original object
?
What would you suggest I call mfcol for these cases?
mfcol <- paste0("Markers", format(Sys.time(), "%a%b%d%H%M%S%Y"))
To be extra safe I can create a new MSnSet i.e. a copy of
object
That's not needed. Just wanted to make sure it's not returned. Not sure about the future; no plans to return at this stage.
Somewhere in the vignette we need a section to explain that the
markers
stored in the feature data can either be amatrix
or acharacter vector
. Default if it's a matrix of markers isMarkers
. SeemrkVecToMat
and friends.