How do I format my data as chi11 in the Get Started documentation?

[liyawang]

I have manually curated my populations in FlowJo.

[rohitfarmer]

Data prepartion for HDStIM from manually gated populations

Hi @liyawang, chi11 is an R list that includes the main input data frame with cells on the rows, markers, and other variables on the columns. The list also contains a vector of state markers, a variable with the column's name with cell populations, a vector of stimulation types, and a variable with the unstim label.

When you say that you want to generate the input data from the manually gated populations from FlowJo, I assume you have FCS files per sample per population exported from FlowJo that you want to read in and generate the input data frame.

Under such conditions, I first prepare a sample information file with the list of FCS files, stimulation types they represent, cell population, subject, etc. Then, I iterate over the FCS files and read in the expression matrix using FlowCore. I attach other variables such as cell population the cells belong to, subject, stimulation type, etc., and concatenate in a single data frame.

Once the data frame is ready, I arcsinh transform the antigen channels. Conventionally, we use a cofactor of 5 for cytof data and 150 for cytek data.

I have uploaded an example dataset to this link https://app.box.com/s/o0gd6605z23bjpxyvl32tybkzqc4iyts with a sample information file. Try the code below to generate the data frame.

library(flowCore)
library(tidyverse)
library(arrow)

data_folder <- file.path("data")
results_folder <- file.path("results")
dir.create(results_folder,recursive = TRUE)

# Load the sample information
cytof_files_fcs <- read_tsv(file.path(data_folder, "sample-information.tsv"),
                            show_col_types = FALSE)

# Iterate over FCS files and concatenate  data into a single data frame
cytof_dat_out <- tibble()
for(j in 1:nrow(cytof_files_fcs)){
        cytof_file_path <- file.path(data_folder, cytof_files_fcs$fcs_file[j])
        cytof_fcs_dat <- read.FCS(cytof_file_path, transformation = FALSE, truncate_max_range = FALSE, alter.names=TRUE)
        cytof_expr_dat <- exprs(cytof_fcs_dat) %>%
                as_tibble()
        cytof_col_names <- as.character(cytof_fcs_dat@parameters@data$name) # Channel names in the FCS columns.
        cytof_col_desc <- as.character(cytof_fcs_dat@parameters@data$desc) # Name of the markers.
        c_names <-  coalesce(cytof_col_desc, cytof_col_names) %>% # User marker names where available else use channel names for example,
        str_split( "_", simplify = TRUE)                          # for time and forward and side scattering.
        c_names <- gsub("-", "_", coalesce(na_if(c_names[,2], ""), c_names[,1]))
        features <- c(gsub("-", "_", panel$antigen), "Time")
        colnames(cytof_expr_dat) <- c_names
        cytof_expr_dat <- dplyr::select(cytof_expr_dat, all_of(features))
        
        # Archsinh transform
        cofactor <- 5
        asi <- function(x) asinh(x/cofactor)
        markers <- setdiff(colnames(cytof_expr_dat), "Time")
        cytof_expr_dat[markers] <- apply(cytof_expr_dat[markers],2, asi)

        cytof_dat_out <- bind_rows(cytof_dat_out, tibble("cell_population" = cytof_files_fcs$cell_population[j],
                                                         "stim_type" = cytof_files_fcs$stim_type[j],
                                                         "subject" = cytof_files_fcs$subject[j],
                                                         cytof_expr_dat))
}

arrow::write_feather(cytof_dat_out, file.path(results_folder, "cytof-dat-asinh-transform.feather"))