server-scripts/qhtcp-workflow/apps/r/join_interaction_zscores.R

suppressMessages({
  library("dplyr")
  library("data.table")
  library("readr")
  library("stringr")
})

# Function to parse arguments
parse_arguments <- function() {
  if (interactive()) {
    args <- c(
      "/home/bryan/documents/develop/hartmanlab/workflow/out/20240116_jhartman2_DoxoHLD",
      3, # sd value
      "/home/bryan/documents/develop/hartmanlab/workflow/out/20240116_jhartman2_DoxoHLD/20240822_jhartman2_DoxoHLD/exp1",
      "Experiment 1: Doxo versus HLD",
      "/home/bryan/documents/develop/hartmanlab/workflow/out/20240116_jhartman2_DoxoHLD/20240822_jhartman2_DoxoHLD/exp2",
      "Experiment 2: HLD versus Doxo"
    )
  } else {
    args <- commandArgs(trailingOnly = TRUE)
  }

  out_dir <- normalizePath(args[1], mustWork = FALSE)
  sd <- as.numeric(args[2])
  paths <- normalizePath(args[seq(3, length(args), by = 2)], mustWork = FALSE)
  names <- args[seq(4, length(args), by = 2)]
  experiments <- setNames(paths, names)

  list(
    out_dir = out_dir,
    sd = sd,
    experiments = experiments
  )
}

args <- parse_arguments()

# Ensure main output directory exists
dir.create(args$out_dir, showWarnings = FALSE, recursive = TRUE)

# Function to read and combine z-score interaction files
combine_zscores <- function(experiments, out_dir) {
  combined_data <- lapply(names(experiments), function(exp_name) {
    exp_dir <- experiments[[exp_name]]
    zscore_file <- file.path(exp_dir, "zscores", "zscore_interactions.csv")

    if (!file.exists(zscore_file)) {
      stop("Z-score file does not exist for ", exp_name, " at ", zscore_file)
    }

    message("Reading z-score file for ", exp_name, " from ", zscore_file)
    data <- fread(zscore_file)
    data$Experiment <- exp_name

    return(data)
  }) %>%
    bind_rows()

  combined_output_file <- file.path(out_dir, "combined_zscores.csv")
  fwrite(combined_data, combined_output_file, row.names = FALSE)

  message("Combined z-score file saved to: ", combined_output_file)
}

# Function to read and combine summary statistics files
combine_summary_stats <- function(experiments, out_dir) {
  combined_stats <- lapply(names(experiments), function(exp_name) {
    exp_dir <- experiments[[exp_name]]
    summary_file <- file.path(exp_dir, "zscores", "summary_stats_all_strains.csv")

    if (!file.exists(summary_file)) {
      stop("Summary stats file does not exist for ", exp_name, " at ", summary_file)
    }

    message("Reading summary stats file for ", exp_name, " from ", summary_file)
    data <- fread(summary_file)
    data$Experiment <- exp_name

    return(data)
  }) %>%
    bind_rows()

  combined_output_file <- file.path(out_dir, "combined_summary_stats.csv")
  fwrite(combined_stats, combined_output_file, row.names = FALSE)

  message("Combined summary stats file saved to: ", combined_output_file)
}

# Function to generate final summary report
generate_final_report <- function(out_dir) {
  combined_zscores <- file.path(out_dir, "combined_zscores.csv")
  combined_stats <- file.path(out_dir, "combined_summary_stats.csv")

  if (!file.exists(combined_zscores) || !file.exists(combined_stats)) {
    stop("Combined z-scores or summary stats files do not exist.")
  }

  zscores_data <- fread(combined_zscores)
  stats_data <- fread(combined_stats)

  message("Merging z-score and summary stats data...")
  final_report <- merge(zscores_data, stats_data, by = c("OrfRep", "Experiment"), all = TRUE, allow.cartesian = TRUE)

  final_report_file <- file.path(out_dir, "final_combined_report.csv")
  fwrite(final_report, final_report_file, row.names = FALSE)

  message("Final combined report saved to: ", final_report_file)
}

# Process all experiments and generate outputs
combine_zscores(args$experiments, args$out_dir)
combine_summary_stats(args$experiments, args$out_dir)
generate_final_report(args$out_dir)