More capitalization corrections and tests reformat for easier reading

R/PNNL_DMS_utils.R

@@ -223,7 +223,7 @@ get_tool_output_files_for_job_number <- function(jobNumber, toolName = NULL,
 {
    # get job records first. This will be useful to get dataset folder
    jobRecord <- get_dms_job_records(jobNumber)
-   datasetFolder <- dirname( as.character(jobRecord$Folder))
+   datasetFolder <- dirname(as.character(jobRecord$folder))
 
    # get tool's subfolder
    if( is.null(toolName) ){
@@ -325,7 +325,7 @@ download_datasets_by_data_package <- function(data_package_num,
    }
    multiproc_cl <- makeCluster(ncores)
    on.exit(stopCluster(multiproc_cl))
-   pbwalk(X = pathToFile$Folder, FUN = file.copy, cl = multiproc_cl, to = copy_to)
+   pbwalk(X = pathToFile$folder, FUN = file.copy, cl = multiproc_cl, to = copy_to)
 }
 
 
@@ -495,7 +495,7 @@ path_to_FASTA_used_by_DMS <- function(data_package_num, organism_db = NULL)
    # make sure it was the same fasta used for all msgf jobs
    # at this point this works only with one data package at a time
    jobRecords <- get_job_records_by_dataset_package(data_package_num)
-   # jobRecords <- jobRecords[grepl("MSGFPlus", jobRecords$Tool),]
+   # jobRecords <- jobRecords[grepl("MSGFPlus", jobRecords$tool),]
    # if(length(unique(jobRecords$organism_db)) != 1){
    #    stop("There should be exactly one FASTA file per data package!")
    # }
@@ -556,6 +556,6 @@ path_to_FASTA_used_by_DMS <- function(data_package_num, organism_db = NULL)
 
 
 # Prevent "no visible binding for global variable" note.
-utils::globalVariables(c("Tool", "value"))
+utils::globalVariables(c("tool", "value"))
 
 

R/read_AScore_results_from_DMS.R

@@ -67,9 +67,9 @@ read_AScore_results_from_DMS <- function(data_package_num){
   }else if(.Platform$OS.type == "windows"){
     # in case Windows
     ascores <- read_tsv(
-      file.path(job['Folder'],"Concatenated_msgfplus_syn_ascore.txt"))
+      file.path(job['folder'],"Concatenated_msgfplus_syn_ascore.txt"))
     job_to_dataset_map <- read_tsv(
-      file.path(job['Folder'], "Job_to_Dataset_Map.txt"))
+      file.path(job['folder'], "Job_to_Dataset_Map.txt"))
   }else{
     stop("unknown OS")
   }
@@ -82,5 +82,5 @@ read_AScore_results_from_DMS <- function(data_package_num){
 
 
 # Prevent "no visible binding for global variable" note.
-utils::globalVariables(c("Tool"))
+utils::globalVariables(c("tool"))
 

R/read_MSstats_from_MSFragger_job.R

@@ -47,18 +47,18 @@ read_MSstats_from_MSFragger_job <- function(data_package_num,
    job_records <- filter(job_records, tool == "MSFragger")
 
    if (!is.null(param_file)) {
-      job_records <- filter(job_records, Parameter_File == param_file)
+      job_records <- filter(job_records, parameter_file == param_file)
    }
 
    if (!is.null(settings_file)) {
-      job_records <- filter(job_records, Settings_File == settings_file)
+      job_records <- filter(job_records, settings_file == !!settings_file)
    }
 
    if (!is.null(organism_db)) {
-      job_records <- filter(job_records, Organism_DB == organism_db)
+      job_records <- filter(job_records, organism_db == !!organism_db)
    }
 
-   path <- unique(job_records$Folder)
+   path <- unique(job_records$folder)
 
    if (length(path) == 0) {
       stop("No jobs found.")

R/read_msfragger_data_from_DMS.R

@@ -60,7 +60,7 @@ read_msfragger_data_from_DMS <- function(data_package_num,
 
    path <- unique(job_records$folder)
    remote_folder <- gsub("\\\\", "/", path)
-   local_folder <- .new_tempdir()
+   mount_folder <- local_folder <- .new_tempdir()
    mount_cmd <- sprintf("mount -t smbfs %s %s", remote_folder, local_folder)
    system(mount_cmd)
 
@@ -82,7 +82,7 @@ read_msfragger_data_from_DMS <- function(data_package_num,
       aggregate_zip_file <- file.path(local_folder, "Dataset_PSM_tsv.zip")
       exdir <- .new_tempdir()
       unzip(zipfile = aggregate_zip_file, list = FALSE, exdir = exdir)
-      path <- exdir
+      local_folder <- exdir
    }
 
    fileNamePttrn <- "_psm\\.tsv"
@@ -108,7 +108,7 @@ read_msfragger_data_from_DMS <- function(data_package_num,
              `Spectrum File` = sub("\\.pep\\.xml", "", `Spectrum File`),
              `Spectrum File` = sub(fileNamePttrn, "", `Spectrum File`))
 
-   system(glue::glue("umount {local_folder}"))
+   system(glue::glue("umount {mount_folder}"))
 
    if (!assume_inference) {
       dt <- dt %>%

R/read_msgf_data_from_DMS.R

@@ -47,11 +47,11 @@ read_msgf_data_from_DMS <- function(data_package_num,
     jobRecords <- get_job_records_by_dataset_package(data_package_num)
   }
 
-  jobRecords <- jobRecords[grepl("MSGFPlus", jobRecords$Tool),]
+  jobRecords <- jobRecords[grepl("MSGFPlus", jobRecords$tool),]
 
   # THIS PARAMETER ARGUMENT CHECK CAN BE MORE ELEGANT
   # Check parameter file. Is there any redundancy?
-  param_files <- unique(jobRecords$Parameter_File)
+  param_files <- unique(jobRecords$parameter_file)
 
   # Check if param_file is NULL
   if (is.null(param_file)) {
@@ -73,7 +73,7 @@ read_msgf_data_from_DMS <- function(data_package_num,
   }
 
   # FASTA files (organism_db)
-  organism_dbs <- unique(jobRecords$`Organism DB`)
+  organism_dbs <- unique(jobRecords$organism_db)
 
   # Check if organism_db is NULL
   if (is.null(organism_db)) {
@@ -95,11 +95,11 @@ read_msgf_data_from_DMS <- function(data_package_num,
   }
 
   # Subset to specific parameter file and FASTA file
-  jobRecords <- jobRecords[jobRecords$Parameter_File == param_file &
-                             jobRecords$`Organism DB` == organism_db, ]
+  jobRecords <- jobRecords[jobRecords$parameter_file == param_file &
+                             jobRecords$organism_db == organism_db, ]
 
   if (use_mzIdentML) {
-    mzid_files <- list.files(jobRecords$Folder, pattern = ".mzid.gz", 
+    mzid_files <- list.files(jobRecords$folder, pattern = ".mzid.gz", 
                              full.names = TRUE)
 
     # Read mzIdentML files into psms slot
@@ -110,7 +110,7 @@ read_msgf_data_from_DMS <- function(data_package_num,
     )
   } else {
     results <- get_results_for_multiple_jobs.dt(jobRecords) 
-    tool <- unique(jobRecords$Tool)
+    tool <- unique(jobRecords$tool)
     pattern <- tool2suffix[[tool]]
     results <- results[[pattern]]
     msnid <- convert_msgf_output_to_msnid(results)

tests/tests.R

@@ -4,23 +4,34 @@ path_check <- function(path) {
     expect_true(grepl("Rtmp", path, ignore.case = FALSE))
     system(glue::glue("umount {path}"))
 }
-
+cat("1------\n")
 expect_true(is_PNNL_DMS_connection_successful())
+cat("2------\n")
 expect_equal(dim(get_dms_job_records(4232)), c(1, 23))
+cat("3------\n")
 expect_equal(get_output_folder_for_job_and_tool(2344924, "DiaNN"), "DNN202408121049_Auto2344924")
+cat("4------\n")
 expect_equal(dim(get_job_records_by_dataset_package(4232)), c(2, 22))
+cat("5------\n")
 expect_equal(dim(get_datasets_by_data_package(4232)), c(1, 13))
+cat("6------\n")
 path <- path_to_FASTA_used_by_DMS(4232)
 path_check(path)
-expect_equal(dim(read_AScore_results_from_DMS(4136)), c(916130, 12)) # check args 
+expect_equal(dim(read_AScore_results_from_DMS(4136)), c(916130, 12)) # check args
+cat("7------\n")
 path <- path_to_study_design_from_DMS(4232)
 path_check(path)
+cat("8------\n")
 path <- path_to_study_design_from_DMS(4232, T)
 path_check(path)
-expect_equal(dim(read_masic_data_from_DMS(4232)), c(20938, 22)) # check args 
+cat("9------\n")
+expect_equal(dim(read_masic_data_from_DMS(4232)), c(20938, 22)) # check args
+cat("10------\n")
 expect_equal(dim(read_msfragger_data_from_DMS(
     4804,
     "MSFragger_Tryp_ProlineRule_ProtNTermAcet_Stat_CysAlk_TMT_6Plex_20ppmParTol.params",
     "MSFragger_PepFDR_0.99_ProtFDR_0.99_IncludeDecoys.xml",
     "ID_008350_56958B5A.fasta"
 )), c(266203, 11))
+
+# get_tool_output_files_for_job_number(2344924, fileNamePttrn = ".") -> res