find_top_features_beta.R

###(Cohort, Assays, Readily, Identified, Now, Algorithm)
#by Gabe & verenna
#******************************************************************************************
#This code finds the top features for a Cohort. Add or take away Tumor types in
#(tumor_type_list) and select how many features to print out by changing the value of (x)
#******************************************************************************************
#/scratch/for_gchavez/aklimate_results/brca/models/
#"/scratch/for_gchavez_old/aklimate_results/",tt,"models"
#the five cohorts we're looking at
tumor_type_list = c("brca", "coadread", "lgggbm", "thym", "ucec")
#for later use
df_cohort = c()
#set x as num of features we want
x = 10
# LOOP A: This for loop will run through all the server cohort files we have
for (i in 1:length(tumor_type_list)) {
  #an abriviation that loops through all five server cohorts
  tt = tumor_type_list[i]
  print(tt)
  #the path to the files
  filepath = paste0("/scratch/for_gchavez_old/aklimate_results/",tt,"/models/")
  # file_name_list: Takes all the files and lists them
  file_name_list = list.files(path = filepath)
  # num_files: The number of files we have
  num_files = length(file_name_list)
  feature_table = c()
  list_of_sets = c()
  #this for loop
  for (j in 1:num_files) {
    file_name = file_name_list[(j)]
    if (length(grep("_multiclass_feature_importance", file_name , fixed = TRUE)) == 1) {
      # Read tabular data into R
      data_table = read.table(
        paste0(filepath,file_name),
        header = TRUE,
        sep = "\t",
        dec = ".",
        stringsAsFactors = FALSE
      )
      # cut data frame to top-x features
      list_of_sets = data_table[c(1:x),]
      # add column with the cross validation fold
      feature_table = rbind(feature_table, list_of_sets)
    }
  }
  print("big list finished")
  print(dim(feature_table))
  #grabs the first 10 rows in our set
  feature_table[order(feature_table$importance, decreasing = TRUE), ]
  top_features = feature_table[1:x, 1]
  top_importances = feature_table[1:x, 2]
  top_list = rbind(top_features, top_importances)
  #initializes an empty vector
  print(top_list)
  print("done")
}