PrepareSpatialDecon error message

[emiliasiukola]

When using the code:
spd <- prepareSpatialDecon(spe_tmm)

we are getting error message:
"NegProbe-WTXmust be included in the dataset. Perhaps specify rmNegProbe=TRUE when using readGeoMx"
when using prepareSpatialDecon.
We are using WTA so there are several negative probes, however we do not understand what the pool= variable in the command should look like. We have already specified rmNegProbe=TRUE when using readGeoMx.

Regards,
Emilia

[ningbioinfo]

Hi @emiliasiukola , the input for the pool argument is a vector with length of the genes, i.e. nrow(spe). It should be something like c("negprob1","negprob1","negprob1","negprob2","negprob2","negprob2"..., "negprobx"), so it tells the function with negative probes is governing which genes.

[kimsjune]

I wonder if the current prepareSpatialDecon function can deal with more than one "NegProbe-WTX". For me, the spe output form readGeoMx with rmNegProbe set to FALSE contains 139 negative probe values with the prefix, NegProbe-WTX_. I'm ignoring normalization for the sake of simplicity. This line in prepareSpatialDecon will return true (shows error) because there is no exact match:

!all("NegProbe-WTX" %in% rownames(spe))
[1] TRUE 

table(grepl("NegProbe-WTX", rownames(spe)))
FALSE TRUE
18676 139

prepareDecon looks similar to derive_GeoMx_background provided by SpatialDecon. I ran the following command instead:

bg <- derive_GeoMx_background(norm = assay(spe, "logcounts"),
probepool = rep(1, nrow(assay(spe, "logcounts"))),
negnames = paste0("NegProbe-WTX_", c(1:139)))

Here, more than one name for negative probes can be set, which I found here. Then I used bg with spatialdecon.

[ningbioinfo]

Hi @kimsjune , yes you're correct, both functions serve the similar purpose. Perhaps I should update it in the next version for automatically do this if multi-probs are detected...