Incorrect intensity threshold due to zero-inflated distribution

[lhollestein]

Dear Yuzhou Feng and team members,

First of all, I would like to congratulate you on the development of this package. It is very user-friendly and has a lot of important functions.

I have a question about the threshold of the intensity.
Due to filtering the intensity of the the tumour marker has a zero-inflated distribution. This leads to problems, as the predict_phenoptype function sets the threshold at 0, making all cells positive for the tumor marker. How can I solve this problem? I would like to find a way to automatically set the threshold the right way, because we need to analyse hundreds of samples, so I would like to avoid setting the threshold manually for each sample.
I copied a few figures and code to explain what's happening:

Distribution of a marker, where threshold setting is ok:

Distribution of the tumor marker:

Amount of zero’s in the tumor marker variable:

table(data_004P1$MelanA==0)
FALSE TRUE
670 157406

Distribution of non-zero values:

Predict phenotype function to set the thresholds:

predict_phenotypes_panel_1<-function(spe_object_name) {
spe_object_name <- predict_phenotypes(spe_object = spe_object_name,
thresholds = NULL,
tumour_marker = "MelanA",
baseline_markers = c("CD8",
"CD79A",
"CD3",
"CD68"),
nuclear_marker = NULL,
reference_phenotypes = FALSE)
return(spe_object_name)
}
SPE_004P1<-predict_phenotypes_panel_1(SPE_004P1)
[1] "CD8 threshold intensity: 145.994177633498"
[1] "CD79A threshold intensity: 212.734985292817"
[1] "CD3 threshold intensity: 143.100134414287"
[1] "CD68 threshold intensity: 1311.24589104864"
[1] "MelanA threshold intensity: 0"

Unique phenotypes, due to incorrect threshold of the tumor marker:

unique(SPE_004P1$Phenotype)
[1] "MelanA" "CD8,MelanA" "CD3,MelanA"
[4] "CD8,CD3,MelanA" "CD79A,MelanA" "CD79A,CD3,MelanA"
[7] "CD8,CD79A,MelanA" "CD79A,CD68,MelanA" "CD68,MelanA"
[10] "CD8,CD68,MelanA" "CD8,CD79A,CD3,MelanA" "CD8,CD79A,CD3,CD68,MelanA"
[13] "CD8,CD3,CD68,MelanA" "CD8,CD79A,CD68,MelanA" "CD3,CD68,MelanA"
[16] "CD79A,CD3,CD68,MelanA"

Could you please advise me how to solve this issue in a way, that it is easy applicable to a large amount of samples?