parse Jackson 2020 data to SingleCellExperiments

jackson-raw-to-sce-server.R

suppressPackageStartupMessages({
  library(tidyverse)
  library(SingleCellExperiment)
  library(argparser)
})

select <- dplyr::select
mutate <- dplyr::mutate
arrange <- dplyr::arrange
rename <- dplyr::rename
filter <- dplyr::filter

## Initial analysis on BaselTMA_SP41_257_X3Y1

write_one_sce <- function(core_test, df_sc, df_loc, output_dir) {
  message(paste("Writting core", core_test))
  df_loc <- filter(df_loc, core == core_test)

  df_count <- df_sc %>%
    filter(core == core_test) %>%
    select(id, channel, mc_counts) %>%
    spread(channel, mc_counts)

  df_count_loc <- inner_join(df_loc, df_count, by = "id")


  ## Parse to an SCE
  ids <- df_count_loc$id

  # bad heuristic -- antibodies have space in name
  df_count_only <- select_at(df_count_loc, vars(contains(" "))) %>%
    as.matrix()
  df_pd <- select_at(df_count_loc, vars(-contains(" "))) %>%
    as.data.frame()

  ## Cell names
  rownames(df_pd) <- rownames(df_count_only) <- ids

  sce <- SingleCellExperiment(
    assays = list(logcounts = t(df_count_only)),
    colData = df_pd
  )

  saveRDS(sce, file.path(output_dir, paste0(core_test, ".RData")))


}

p <- arg_parser("Read Jackson 2020 data")

p <- add_argument(p, "--input_sc", "Input SC_dat.csv")
p <- add_argument(p, "--input_loc", "Input location csv")
p <- add_argument(p, "--output_dir", "Output directory")

argv <- parse_args(p)

df_sc <- read_csv(argv$input_sc)
df_loc <- read_csv(argv$input_loc)

cores <- sort(unique(df_sc$core))

for(core in cores) {
  write_one_sce(core, df_sc, df_loc, argv$output_dir)
}