.Rhistory

cs <- load_cytoset_from_fcs(path = "../data/fcs_data/",pattern = "TNK-CR1")
# add mock metadata
set.seed(123)
meta <- data.frame(name = sampleNames(cs),
status = "Healthy",
panel = "T Cell",
mock_treatment = sample(x = c("Treated","Control"),size = 4,prob = c(0.5,0.5),replace = TRUE),
row.names = sampleNames(cs))
pData(cs) <- meta
# creating a GatingSet
gs <- flowWorkspace::GatingSet(cs)
spill <- keyword(gs[[1]],"$SPILLOVER") # extract spillover matrix stored within the file
compensate(gs,spill) # GatingSet will be compensated and stores the spill matrix as well
recompute(gs) # update the gs
my_trans <- readRDS(file = "../data/fj_wsp/fj_transform")
# make into transformerList object
my_trans_list <- flowWorkspace::transformerList(from = names(my_trans),
trans = my_trans)
gs <- flowWorkspace::transform(gs, my_trans_list) # transforms underlying data
# calculate and add a singlet gate
singlet_gate <- fsApply(gs_pop_get_data(gs),function(x){ # sample wise calculation using fsApply
flowStats::gate_singlet(x,
filterId = "singlet", # name for the gate
wider_gate = FALSE, # indicate if the returned gate should be linient
prediction_level = 0.95,
subsample_pct = 0.5 # percent of events to sample to calculate the gate
)
})
# add gate to the GatingSet
gs_pop_add(gs,singlet_gate, parent = "root") # add singlet_gate to the root node
recompute(gs) # recompute updates the gs
# calculate a live gate
## Example of a quantile gate
live_gate <- fsApply(gs_pop_get_data(gs,"singlet"), # indicating that estimation should be done on filtered data: singlet
function(x){
gate <- openCyto::gate_quantile(fr = x, channel = "U450-A",
probs = 0.95,
filterId = "live"
)
# keep negative events
gate@max <- gate@min
gate@min <- -Inf
return(gate)
})
# add live gate
gs_pop_add(gs, parent = "singlet",gate = live_gate)
recompute(gs)
# calculate lymphocyte gate
## Example of using flowClust
lymphocyte_gate <- fsApply(gs_pop_get_data(gs,"live"),
function(x){
openCyto::gate_flowclust_2d(fr = x,
xChannel = "FSC-A",
yChannel = "SSC-A",
filterId = "lymphocytes",
K = 1,target = c(1E5,0.5E3))
})
# call it lymphocytes
lymphocyte_gate <- lapply(lymphocyte_gate,
function(x){
x@filterId <- "lymphocytes"
x
})
# add lymphocyte gate
gs_pop_add(gs, parent = "live", gate = lymphocyte_gate)
recompute(gs)
# add T cells gate
## Example of rectangleGate
# using rectangle gate to add T cell gate
cd3_rectanlge <- matrix(c(140, 205, 0, 200),
nrow = 2,ncol = 2,
byrow = F,
dimnames = list(NULL,c("V510-A","U570-A"))) # channel names
cd3_rectangle_gate <- rectangleGate(.gate = cd3_rectanlge,filterId = "CD3+ T cells")
# add gate
gs_pop_add(gs, gate = cd3_rectangle_gate, parent = "lymphocytes")
recompute(gs)
# add NKT cell gate
## Example of polygonGate
# define coordinates
nkt_poly <- matrix(c(
115,140,
150,150,
150,180,
200,180,
200,140), # coordinates
ncol = 2,
byrow = T,
dimnames = list(NULL, c("R670-A","V510-A"))
)
# convert to gate
nkt_poly_gate <- polygonGate(nkt_poly,
filterId = "NKT cells")
# move gate and make smaller
nkt_poly_gate <- flowCore::transform_gate(nkt_poly_gate,
dx = 1,
scale = c(1.05,1.05)
)
# add to gs
gs_pop_add(gs,nkt_poly_gate, parent = "CD3+ T cells")
recompute(gs)
# Add non-NKT cells
## Example rangeGate
# make a list of sample specific gates
non_nkt <- lapply(gs,
function(x){
ff <- gh_pop_get_data(x, "CD3+ T cells") # extract data at specific node for cleaner calculation
flowStats::rangeGate(ff,
stain = "R670-A",
filterId = "non-NKT Cells",
positive = FALSE,
alpha = 0.1
)
})
non_nkt$`1615fa39c8b_4002_TNK-CR1.fcs` <- transform_gate(non_nkt$`1615fa39c8b_4002_TNK-CR1.fcs`,
dx = 25)
plot(gs)
non_nkt
transform_gate(non_nkt$`1615fa39c8b_4002_TNK-CR1.fcs`,
dx = 25)
non_nkt$`1615fa39c8b_4002_TNK-CR1.fcs` <- transform_gate(non_nkt$`4002_TNK-CR1.fcs`,
dx = 25)
# add non_nkt
gs_pop_add(gs, non_nkt,parent = "CD3+ T cells")
sampleNames(gs)
# make a list of sample specific gates
non_nkt <- lapply(gs,
function(x){
ff <- gh_pop_get_data(x, "CD3+ T cells") # extract data at specific node for cleaner calculation
flowStats::rangeGate(ff,
stain = "R670-A",
filterId = "non-NKT Cells",
positive = FALSE,
alpha = 0.1
)
})
non_nkt[[2]] <- transform_gate(non_nkt[[2]],
dx = 25)
# add non_nkt
gs_pop_add(gs, non_nkt,parent = "CD3+ T cells")
recompute(gs)
# identify conventional T cells
# Example use multiple gating tools and piping
## Estimate gate by sampling the gatinset
conv_t_cell_gate <- gs |>
gs_pop_get_data(y = "non-NKT Cells") |>
cytoset_to_flowSet() |>
(function(x){
set.seed(123)
sample_n <- 1E3
all_exprs <- fsApply(x,function(y)exprs(y)[sample(nrow(y),sample_n),])
all_exprs <- flowFrame(all_exprs)
attrs = c("min","max")
# identify cutpoints
g1 <- openCyto::gate_quantile(all_exprs,
channel = "B515-A",
prob = 0.98
)
g1_attrs <- sapply(attrs, function(y){
attr(g1,y)
})
g2 <- openCyto::gate_quantile(all_exprs,
channel = "G575-A",
prob = 0.90
)
g2_attrs <- sapply(attrs, function(y){
attr(g2,y)
})
# get cutpoints
g1_cutpoint <- g1_attrs[!is.infinite(g1_attrs)]
g2_cutpoint <- g2_attrs[!is.infinite(g2_attrs)]
# make rectangleGate
qg <- rectangleGate(list("B515-A" = c(g1_cutpoint,-Inf),
"G575-A" = c(g2_cutpoint,-Inf)),
filterId = "conv_Tcells")
})()
# adjust gates
conv_t_cell_gate <- flowCore::transform_gate(conv_t_cell_gate,
dx = 11,
dy = 11
)
# add quad gate
gs_pop_add(gs, conv_t_cell_gate,parent = "non-NKT Cells")
recompute(gs)
# identify MAIT cells
# add MAIT cells
## Example of estimation using collapsed data
mait_gate_all <- openCyto::gate_flowclust_2d(fr =  flowFrame(fsApply(gs_pop_get_data(gs,"conv_Tcells"),
function(y){
sample_n = min(nrow(y), 1E4)
exprs(y)[sample(nrow(y),sample_n),]
}
)
),
xChannel = "V710-A",
yChannel = "G660-A",target = c(175,175),
K = 14,filterId = "MAIT Cells" # K indicates number of clusters to estimate
)
# fix filterId
mait_gate_all@filterId <- "MAIT Cells"
# scale gate
mait_gate_all <- scale_gate(mait_gate_all, scale = 2)
# add gate
gs_pop_add(gs,gate = mait_gate_all, parent = "conv_Tcells")
recompute(gs)
# add not MAIT gate
## Example of booleanFilter
not_mait <- booleanFilter(`!MAIT Cells`, filterId = "not_MAIT")
# add boolean gate
gs_pop_add(gs, not_mait, parent = "conv_Tcells")
Example of how to use booleanFilter to create a polygon gate
# Example of how to use booleanFilter to create a polygon gate
not_mait_gate <- lapply(seq(1,length(
gs_pop_get_gate(gs,"MAIT Cells")),1),
function(x){
gate = gh_pop_get_gate(gs[[x]],"MAIT Cells")
# enlarge for stringency
gate <- scale_gate(gate,3)
ff <- cytoframe_to_flowFrame(gh_pop_get_data(gs[[x]],"conv_Tcells"))
fl <- filter(ff,gate)
idx <- which(!fl@subSet)
set.seed(123)
sample_n <- min(length(idx),1E5)
m.names <- colnames(gate@cov)
events <- exprs(ff)[idx[sample(length(idx),sample_n)],m.names]
c.hull <- chull(events)
polygon_mait <- polygonGate(.gate = events[
c(c.hull,c.hull[1]),],
filterId = "not_MAIT_Polygon")
}
);names(not_mait_gate) <- sampleNames(gs)
# add gate
gs_pop_add(gs,not_mait_gate,parent = "conv_Tcells")
recompute(gs)
# add a quad gate
cd4_cd8_quad_gate <- lapply(gs_pop_get_data(gs,"not_MAIT"),
function(x){
qg <- openCyto::gate_quad_tmix(x,
channels = c("U785-A",
"V570-A"),
K = 2)
# give readable names
r.names <- c("U785-A" = "CD4", # these are human readable names
"V570-A" = "CD8a")
for(i in 1:length(qg)){ # iterate over each quadrant
sapply(1:length(r.names),function(x){ # iterate over human readable names
current.name <- attr(qg[[i]],"filterId") # keep a record of the current name
attr(qg[[i]],"filterId") <<- gsub( # substitute and apply to qg
pattern = names(r.names)[x],
replacement = r.names[x],
x = current.name)
})
}
return(qg) # return modified gate
})
# add gate
gs_pop_add(gs, cd4_cd8_quad_gate,parent = "not_MAIT_Polygon")
recompute(gs)
# cleanup
rm(list = ls()[!(ls() %in% "gs")])
autoplot(gs[[1]])
library(ggcyto)
autoplot(gs[[1]])
autoplot(gs[[1]], inverse.transform = T)
rm(gs)
gs <- "a"
rm(gs)
cs <- load_cytoset_from_fcs("../data/fcs_data/",pattern = "TNK-CR1")
cs <- load_cytoset_from_fcs(path = "../data/fcs_data/",pattern = "TNK-CR1")
# add mock metadata
set.seed(123)
meta <- data.frame(name = sampleNames(cs),
status = "Healthy",
panel = "T Cell",
mock_treatment = sample(x = c("Treated","Control"),size = 4,prob = c(0.5,0.5),replace = TRUE),
row.names = sampleNames(cs))
pData(cs) <- meta
# creating a GatingSet
gs <- flowWorkspace::GatingSet(cs)
spill <- keyword(gs[[1]],"$SPILLOVER") # extract spillover matrix stored within the file
compensate(gs,spill) # GatingSet will be compensated and stores the spill matrix as well
recompute(gs) # update the gs
my_trans <- readRDS(file = "../data/fj_wsp/fj_transform")
# make into transformerList object
my_trans_list <- flowWorkspace::transformerList(from = names(my_trans),
trans = my_trans)
gs <- flowWorkspace::transform(gs, my_trans_list) # transforms underlying data
gh_get_transformations(gs[[1]])
singlet_gate <- fsApply(gs_pop_get_data(gs),function(x){ # sample wise calculation using fsApply
flowStats::gate_singlet(x,
filterId = "singlet", # name for the gate
wider_gate = FALSE, # indicate if the returned gate should be linient
prediction_level = 0.95,
subsample_pct = 0.5 # percent of events to sample to calculate the gate
)
})
# add gate to the GatingSet
gs_pop_add(gs,singlet_gate, parent = "root") # add singlet_gate to the root node
recompute(gs) # recompute updates the gs
# calculate a live gate
## Example of a quantile gate
live_gate <- fsApply(gs_pop_get_data(gs,"singlet"), # indicating that estimation should be done on filtered data: singlet
function(x){
gate <- openCyto::gate_quantile(fr = x, channel = "U450-A",
probs = 0.95,
filterId = "live"
)
# keep negative events
gate@max <- gate@min
gate@min <- -Inf
return(gate)
})
# add live gate
gs_pop_add(gs, parent = "singlet",gate = live_gate)
recompute(gs)
# calculate lymphocyte gate
## Example of using flowClust
lymphocyte_gate <- fsApply(gs_pop_get_data(gs,"live"),
function(x){
openCyto::gate_flowclust_2d(fr = x,
xChannel = "FSC-A",
yChannel = "SSC-A",
filterId = "lymphocytes",
K = 1,target = c(1E5,0.5E3))
})
# call it lymphocytes
lymphocyte_gate <- lapply(lymphocyte_gate,
function(x){
x@filterId <- "lymphocytes"
x
})
# add lymphocyte gate
gs_pop_add(gs, parent = "live", gate = lymphocyte_gate)
recompute(gs)
# add T cells gate
## Example of rectangleGate
# using rectangle gate to add T cell gate
cd3_rectanlge <- matrix(c(140, 205, 0, 200),
nrow = 2,ncol = 2,
byrow = F,
dimnames = list(NULL,c("V510-A","U570-A"))) # channel names
cd3_rectangle_gate <- rectangleGate(.gate = cd3_rectanlge,filterId = "CD3+ T cells")
# add gate
gs_pop_add(gs, gate = cd3_rectangle_gate, parent = "lymphocytes")
recompute(gs)
# add NKT cell gate
## Example of polygonGate
# define coordinates
nkt_poly <- matrix(c(
115,140,
150,150,
150,180,
200,180,
200,140), # coordinates
ncol = 2,
byrow = T,
dimnames = list(NULL, c("R670-A","V510-A"))
)
# convert to gate
nkt_poly_gate <- polygonGate(nkt_poly,
filterId = "NKT cells")
# move gate and make smaller
nkt_poly_gate <- flowCore::transform_gate(nkt_poly_gate,
dx = 1,
scale = c(1.05,1.05)
)
# add to gs
gs_pop_add(gs,nkt_poly_gate, parent = "CD3+ T cells")
recompute(gs)
# Add non-NKT cells
## Example rangeGate
# make a list of sample specific gates
non_nkt <- lapply(gs,
function(x){
ff <- gh_pop_get_data(x, "CD3+ T cells") # extract data at specific node for cleaner calculation
flowStats::rangeGate(ff,
stain = "R670-A",
filterId = "non-NKT Cells",
positive = FALSE,
alpha = 0.1
)
})
non_nkt[[2]] <- transform_gate(non_nkt[[2]],
dx = 25)
# add non_nkt
gs_pop_add(gs, non_nkt,parent = "CD3+ T cells")
recompute(gs)
# identify conventional T cells
# Example use multiple gating tools and piping
## Estimate gate by sampling the gatinset
conv_t_cell_gate <- gs |>
gs_pop_get_data(y = "non-NKT Cells") |>
cytoset_to_flowSet() |>
(function(x){
set.seed(123)
sample_n <- 1E3
all_exprs <- fsApply(x,function(y)exprs(y)[sample(nrow(y),sample_n),])
all_exprs <- flowFrame(all_exprs)
attrs = c("min","max")
# identify cutpoints
g1 <- openCyto::gate_quantile(all_exprs,
channel = "B515-A",
prob = 0.98
)
g1_attrs <- sapply(attrs, function(y){
attr(g1,y)
})
g2 <- openCyto::gate_quantile(all_exprs,
channel = "G575-A",
prob = 0.90
)
g2_attrs <- sapply(attrs, function(y){
attr(g2,y)
})
# get cutpoints
g1_cutpoint <- g1_attrs[!is.infinite(g1_attrs)]
g2_cutpoint <- g2_attrs[!is.infinite(g2_attrs)]
# make rectangleGate
qg <- rectangleGate(list("B515-A" = c(g1_cutpoint,-Inf),
"G575-A" = c(g2_cutpoint,-Inf)),
filterId = "conv_Tcells")
})()
# adjust gates
conv_t_cell_gate <- flowCore::transform_gate(conv_t_cell_gate,
dx = 11,
dy = 11
)
# add quad gate
gs_pop_add(gs, conv_t_cell_gate,parent = "non-NKT Cells")
recompute(gs)
# identify MAIT cells
# add MAIT cells
## Example of estimation using collapsed data
mait_gate_all <- openCyto::gate_flowclust_2d(fr =  flowFrame(fsApply(gs_pop_get_data(gs,"conv_Tcells"),
function(y){
sample_n = min(nrow(y), 1E4)
exprs(y)[sample(nrow(y),sample_n),]
}
)
),
xChannel = "V710-A",
yChannel = "G660-A",target = c(175,175),
K = 14,filterId = "MAIT Cells" # K indicates number of clusters to estimate
)
# fix filterId
mait_gate_all@filterId <- "MAIT Cells"
# scale gate
mait_gate_all <- scale_gate(mait_gate_all, scale = 2)
# add gate
gs_pop_add(gs,gate = mait_gate_all, parent = "conv_Tcells")
recompute(gs)
# add not MAIT gate
## Example of booleanFilter
not_mait <- booleanFilter(`!MAIT Cells`, filterId = "not_MAIT")
# add boolean gate
gs_pop_add(gs, not_mait, parent = "conv_Tcells")
recompute(gs)
# Example of how to use booleanFilter to create a polygon gate
not_mait_gate <- lapply(seq(1,length(
gs_pop_get_gate(gs,"MAIT Cells")),1),
function(x){
gate = gh_pop_get_gate(gs[[x]],"MAIT Cells")
# enlarge for stringency
gate <- scale_gate(gate,3)
ff <- cytoframe_to_flowFrame(gh_pop_get_data(gs[[x]],"conv_Tcells"))
fl <- filter(ff,gate)
idx <- which(!fl@subSet)
set.seed(123)
sample_n <- min(length(idx),1E5)
m.names <- colnames(gate@cov)
events <- exprs(ff)[idx[sample(length(idx),sample_n)],m.names]
c.hull <- chull(events)
polygon_mait <- polygonGate(.gate = events[
c(c.hull,c.hull[1]),],
filterId = "not_MAIT_Polygon")
}
);names(not_mait_gate) <- sampleNames(gs)
# add gate
gs_pop_add(gs,not_mait_gate,parent = "conv_Tcells")
recompute(gs)
# add a quad gate
cd4_cd8_quad_gate <- lapply(gs_pop_get_data(gs,"not_MAIT"),
function(x){
qg <- openCyto::gate_quad_tmix(x,
channels = c("U785-A",
"V570-A"),
K = 2)
# give readable names
r.names <- c("U785-A" = "CD4", # these are human readable names
"V570-A" = "CD8a")
for(i in 1:length(qg)){ # iterate over each quadrant
sapply(1:length(r.names),function(x){ # iterate over human readable names
current.name <- attr(qg[[i]],"filterId") # keep a record of the current name
attr(qg[[i]],"filterId") <<- gsub( # substitute and apply to qg
pattern = names(r.names)[x],
replacement = r.names[x],
x = current.name)
})
}
return(qg) # return modified gate
})
# add gate
gs_pop_add(gs, cd4_cd8_quad_gate,parent = "not_MAIT_Polygon")
recompute(gs)
gh_get_transformations(gs[[1]])
autoplot(gs[[1]], inverse.transform = T)
?autoplot
autoplot(gs[[1]], axis_inverse_trans = T)
gh_get_transformations(gs[[1]])
autoplot(gs[[1]], gate = "lymphocytes", axis_inverse_trans = TRUE)
autoplot(gs[[1]], gate = "lymphocytes", axis_inverse_trans = FALSE)
autoplot(gs[[1]], gate = "lymphocytes")+axis_y_inverse_trans()