Visualise ROI masks in v3.6

[Matthew Smith]

Hello,

I am struggling to visualise my selected ROIs as masks over my image with v3.6.

With the previous version I used:

levels = makeFactor(L1 = ..., L2 = ..., ...)

image(MSIobject, levels ~ x * y, key=T)

However, I now get the error:

Error in rep.int(list(X), nrow(vals)) : invalid type (NULL) for 'times' (must be a vector)

If I attach it to the spectra data I can generate the masks, but I lose the factor names

spectra(cal_mix, "levels") = t(levels)

Is there a better way to do this?

Best wishes,

Matt

[kbemis]

Add the variable as a column to the pixelData.

pData(MSIobject)$levels <- levels

image(MSIobject, "levels") # simplest way

image(MSIobject, levels ~ x * y, i=NULL) # alternative way

These should work. Passing just a character vector as the 2nd argument will look for the column(s) in pixelData. If using a formula, setting i=NULL says to look in pixelData rather than spectraData (where i is used to resolve the rows of the intensity matrix).

The code for correctly resolving names in the formula that aren't part of the object were very subject to breaking, so 3.6 now has stricter rules for resolving these names as above.

You can find an example here:

https://bioconductor.org/packages/release/bioc/vignettes/Cardinal/inst/doc/Cardinal3-guide.html#summarization-e.g.-mean-spectra-tic-etc.

-Kylie

[Matthew Smith]

Perfect, that has resolved it. Thanks :)