warpset cutting off low-end or high end data

[The1stMartian]

Hi, I'm having an issue where warpset is changing values at the lowest end of certain samples, making them a single (lower) value. What is the cause of this and how do I prevent it from happening?

To see if some of the lower values were simply below some threshold, I added a single value to every cell in every channel, effectively shifting everything to the right, then I did the normalization again. This time the lower values were spared, while the upper values were cut off and moved to a single max value. Again, I'm not sure what I need to adjust to avoid this.

The command I've been using is "flowStats::warpSet(x = fixed, stains = fluorophores2, clipRange = 0)". (I set clipRange to zero thinking that this option could be to blame, but I essentially see the same thing when I leave it as default.)

[mikejiang]

reproducible example script with data, please

[The1stMartian]

reproducible example script with data, please

Hi Mike, I'm afraid I can't upload the data as it's not mine, but here's the script.

inputFolder = "/input"
outputFolder = "/output"
flowSetFolder = "/FlowSet"
metaDataFile = "FCMetadata.csv"
normFolder = "NormalizationData"

suppressMessages(library(tidyverse))
suppressMessages(library(flowCore))
suppressMessages(library(flowAI))
suppressMessages(library(flowAssist))

dir.create(file.path(outputFolder, flowSetFolder), recursive = TRUE, showWarnings = FALSE)
dir.create(file.path(outputFolder, normFolder), recursive = TRUE, showWarnings = FALSE)

myfiles <- list.files(path=inputFolder, pattern = ".FCS", ignore.case = TRUE)
input.FCS <- read.flowSet(myfiles, path=inputFolder, truncate_max_range = FALSE)

input.FCS.comp <-compensate(input.FCS, spillover(input.FCS[[1]])$SPILL)
rm(input.FCS)

input.FCS.comp.cleaned <- flow_auto_qc(input.FCS.comp)
rm(input.FCS.comp)

trans <- estimateLogicle(input.FCS.comp.cleaned[[1]], fluorophores)
input.FCS.comp.cleaned.trans <- transform(input.FCS.comp.cleaned, trans)
rm(input.FCS.comp.cleaned)

removeList = c("FSC-A", "FSC-H", "FSC-W", "SSC-A", "SSC-H","SSC-W", "Time")
fluorophores<-as.character(colnames(exprs(input.FCS.comp.cleaned.trans[[1]])))
fluorophores<-fluorophores[!(fluorophores %in% removeList)]

normalizedData = flowStats::warpSet(x = input.FCS.comp.cleaned.trans, stains = fluorophores, clipRange = 0)

png(file.path(outputFolder, normFolder, "FluorDensity_Pre.png"), width = 1600, height = 1600)
densityplot(~ ., input.FCS.comp.cleaned.trans)
dev.off()

png(file.path(outputFolder, normFolder, "FluorDensity_Post.png"), width = 1600, height = 1600)
densityplot(~ ., normalizedData)
dev.off()

multi_gs <- GatingSet(normalizedData)

marker1 = "FSC-A"
protein1 = "Forward Scatter"
marker2 = "SSC-A"
protein2 = "Side Scatter"
currentPopName = "non_debris"

fs_data <- gs_pop_get_data(multi_gs)

#non_debris <- fsApply(fs_data, function(fr) openCyto:::.flowClust.2d(fr, channels = c("FSC-A","SSC-A")))
non_debris <- rectangleGate(filterId="Non_debris_manual", "FSC-A" = c(0.25E5,3E5), "SSC-A" = c(10,3E5))

gs_pop_add(multi_gs, non_debris, parent = "root", name="non_debris")
#gs_pop_remove(multi_gs, "non_debris")
recompute(multi_gs)

plt = autoplot(multi_gs, x=marker1, y=marker2, currentPopName, bins=2012) +
labs(x= marker1, y = marker2) + theme(text=element_text(size = 16))
plt_path = file.path(inputFolder, gatingFolder, paste0(currentPopName,".png"))
ggsave(plt_path, plt, width = 30, height = 30)
plt

marker1 = "SSC-A"
marker2 = "SSC-H"
currentPopName = "singlets"
parentPopName = "non_debris"
currentPopName = "singlets"

fs_data <- gs_pop_get_data(multi_gs, parentPopName)
singlets <- fsApply(fs_data, function(fr) openCyto:::.singletGate(fr, channels =c(marker1, marker2)))
gs_pop_add(multi_gs, singlets, parent = parentPopName, name = currentPopName)
recompute(multi_gs)

plt = autoplot(multi_gs, x = marker1, y = marker2, currentPopName, bins = 256) +labs(x= marker1, y = marker2)
plt_path = file.path(outputFolder, paste0(currentPopName, ".png"))
ggsave(plt_path, plt, width = 30, height = 30)
print("Finished saving singlets plot...")
plt

marker1 = "SSC-A"
protein1 = "Side Scatter"
marker2 = "BV570-A"
protein2 = "Vital Stain"
currentPopName = "live"
parentPopName = "singlets"

fs_data <- gs_pop_get_data(multi_gs, parentPopName)

x_boundary = c(10,2E5)
y_boundary = c(0,3.5)
live <- rectangleGate(filterId="livegateID", "SSC-A" = x_boundary, "BV570-A" = y_boundary)

gs_pop_add(multi_gs, live, parent = parentPopName, name=currentPopName)

recompute(multi_gs)

plt = autoplot(multi_gs, x=marker1, y=marker2, currentPopName, bins=512) +
labs(x = paste(marker1, protein1), y = paste(marker2, protein2)) +
theme(text=element_text(size = 16))
plt_path = file.path(outputFolder, paste0(currentPopName,".png"))
ggsave(plt_path, plt, width = 30, height = 30)
print("Finished saving live/dead plot...")

marker1 = "BV570-A"
protein1 = "Vital Stain"
marker2 = "BUV496-A"
protein2 = "CD45"
parentPopName = "live"
currentPopName = "CD45+"

fs_data <- gs_pop_get_data(multi_gs, parentPopName)

x_boundary = c(1, 4)
y_boundary = c(2.5,5)
CD45pos <- rectangleGate(filterId="CD45", "BV570-A" = x_boundary, "BUV496-A" = y_boundary)

gs_pop_add(multi_gs, CD45pos, parent = parentPopName, name=currentPopName)
recompute(multi_gs)

plt = autoplot(multi_gs, x=marker1, y=marker2, currentPopName, bins=512) +
ggcyto_par_set(limits = list(x = c(0, 5), y= c(0, 5))) +
labs(x = paste(marker1, protein1, "\n\nPopulation:", currentPopName), y = paste(marker2, protein2)) +
theme(text=element_text(size = 16))
plt_path = file.path(outputFolder, paste0(currentPopName,".png"))
ggsave(plt_path, plt, width = 30, height = 30)
print("Finished saving CD45 plot...")

marker1 = "BUV805-A"
protein1 = "CD16"
marker2 = "BUV395-A"
protein2 = "CD56"
currentPopName = "CD16+"
parentPopName = "CD45+"

CD16pos <- rectangleGate(filterId="CD16_neg", "BUV805-A" = c(2, 6), "BUV395-A" = c(0,5))

gs_pop_add(multi_gs, CD16pos, parent = parentPopName, name=currentPopName)
recompute(multi_gs)

plt = autoplot(multi_gs, x=marker1, y=marker2, currentPopName, bins=512) +
ggcyto_par_set(limits = list(x = c(0, 9), y= c(0, 9))) +
labs(x = paste(marker1, protein1, "\n\nPopulation:", currentPopName), y = paste(marker2, protein2)) +
theme(text=element_text(size = 16))
plt_path = file.path(outputFolder, paste0(currentPopName,".png"))
ggsave(plt_path, plt, width = 30, height = 30)