init read_FragPipe_LFQ and read_FragPipe_TMT

DESCRIPTION

@@ -90,4 +90,4 @@ Remotes:
     github::cran/FField,
     github::PNNL-Comp-Mass-Spec/MSnID@pnnl-master
 VignetteBuilder: knitr
-RoxygenNote: 7.2.3
+RoxygenNote: 7.3.1

NAMESPACE

@@ -57,6 +57,8 @@ export(readMaxQuantPeptides)
 export(readMaxQuantProtGroups)
 export(readMaxQuantSummary)
 export(readSkyLinePRR)
+export(read_FragPipe_LFQ)
+export(read_FragPipe_TMT)
 export(remove_batch_effect)
 export(remove_covariate)
 export(rf_modeling)
@@ -106,6 +108,7 @@ importFrom(ROCR,prediction)
 importFrom(WGCNA,empiricalBayesLM)
 importFrom(car,Anova)
 importFrom(circlize,colorRamp2)
+importFrom(data.table,fread)
 importFrom(data.table,rbindlist)
 importFrom(data.table,setDT)
 importFrom(doParallel,registerDoParallel)
@@ -118,6 +121,7 @@ importFrom(dplyr,case_when)
 importFrom(dplyr,desc)
 importFrom(dplyr,distinct)
 importFrom(dplyr,do)
+importFrom(dplyr,everything)
 importFrom(dplyr,filter)
 importFrom(dplyr,full_join)
 importFrom(dplyr,group_by)
@@ -305,6 +309,7 @@ importFrom(stringr,str_locate_all)
 importFrom(stringr,str_replace_all)
 importFrom(stringr,str_sub)
 importFrom(sva,ComBat)
+importFrom(tibble,column_to_rownames)
 importFrom(tibble,enframe)
 importFrom(tibble,rownames_to_column)
 importFrom(tidyr,fill)

R/read_FragPipe_LFQ.R

@@ -0,0 +1,71 @@
+#' Reading MSFragger-generated LFQ-based MSstats from a file path as MSnSet object
+#'
+#' @description Function has only been tested with label-free intensity-based
+#'   quantification data. MSstats.csv is
+#'   an optional output file which needs to be specified in FP settings.
+#'
+#' @param path character; File path to the FragPipe-generated MSstats.csv file
+#'
+#' @return (MSnSet) MSnSet object of MSFragger LFQ results
+#'
+#' @importFrom MSnbase MSnSet
+#' @importFrom data.table fread
+#' @importFrom tidyr pivot_wider
+#' @importFrom tibble column_to_rownames
+#' @importFrom dplyr %>% select filter distinct relocate everything mutate
+#'
+#' @examples
+#' file_path <- "C:/Users/fakeusr222/Desktop/MSF_LFQ_job/MSstats.csv"
+#'   msnset <- read_FragPipe_LFQ(file_path)
+#'   show(msnset)
+#'
+#' @export read_FragPipe_LFQ
+
+
+read_FragPipe_LFQ <- function(path = NULL)
+{
+  path_to_file <- path
+
+  if (!file.exists(path_to_file)) {
+    stop(sprintf("MSstats.csv file not found in folder: %s", dirname(path_to_file)))
+  }
+
+  df <- fread(file = path_to_file, showProgress = FALSE, data.table = FALSE) %>%
+    filter(!is.na(Intensity)) %>%
+    # May add charge col later
+    select(ProteinName, PeptideSequence, Run, Intensity) %>%
+    mutate(featureName = paste0(ProteinName, "@", PeptideSequence)) %>%
+    relocate(featureName, .before = everything())
+
+  # Will sum intensity of unique features.
+  x_data <- df %>%
+    pivot_wider(id_cols = "featureName",
+                names_from = "Run",
+                values_from = "Intensity",
+                values_fn = sum) %>%
+    as.data.frame() %>%
+    column_to_rownames(var = "featureName") %>%
+    as.matrix()
+
+  f_data <- df %>%
+    distinct(featureName, ProteinName, PeptideSequence) %>%
+    `rownames<-`(.[["featureName"]])
+
+  p_data <- df %>%
+    distinct(Run) %>%
+    `rownames<-`(.[["Run"]])
+
+  x_data <- x_data[rownames(f_data), rownames(p_data)]
+
+  m <- MSnSet(exprs = x_data, fData = f_data, pData = p_data)
+
+  return(m)
+}
+
+
+
+
+utils::globalVariables(
+  c("ProteinName", "PeptideSequence", "Run", "Intensity", ".", "featureName")
+)
+

R/read_FragPipe_TMT.R

@@ -0,0 +1,70 @@
+#' Reading MSFragger-generated tmt-report files from a file path as MSnSet object
+#'
+#' @description Function has only been tested with TMT intensity-based
+#'   quantification data. Desired tmt-report output files (e.g., "ratio_multi-site_MD.tsv")
+#'   must be properly selected in the FP settings.
+#'
+#' @param path character; File path to the desired FragPipe-generated tmt-report file.
+#'   Any tmt-report file may be used.
+#'
+#' @return (MSnSet) MSnSet object of MSFragger TMT results
+#'
+#' @importFrom MSnbase MSnSet
+#' @importFrom data.table fread
+#' @importFrom tibble column_to_rownames
+#' @importFrom dplyr %>% select mutate
+#'
+#' @examples
+#' file_path <- "C:/Users/fakeusr222/Desktop/MSF_TMT_job/ratio_multi-site_MD.tsv"
+#'   msnset <- read_FragPipe_TMT(file_path)
+#'   show(msnset)
+#'
+#' @export read_FragPipe_TMT
+
+read_FragPipe_TMT <- function(path = NULL)
+{
+
+  path_to_file <- path
+
+  if (!file.exists(path_to_file)) {
+    stop(sprintf("file not found in folder: %s", dirname(path_to_file)))
+  }
+
+  df <- fread(file = path_to_file, showProgress = FALSE, data.table = FALSE)
+
+
+  # make featureNames
+  if (grepl("multi-site|single-site|peptide", basename(path_to_file))) {
+    df <- df %>%
+      mutate(rowname = paste(Gene, ProteinID, Peptide, sep = "|"))
+  } else if (grepl("gene", basename(path_to_file))) {
+    df <- df %>%
+      mutate(rowname = paste(Index, ProteinID, sep = "|"))
+  } else if (grepl("protein", basename(path_to_file))) {
+    df <- df %>%
+      mutate(rowname = paste(Gene, Index, sep = "|"))
+  }
+
+  df <- df %>%
+    mutate(featureName = rowname, .before = colnames(.)[[1]]) %>%
+    column_to_rownames(var = "rowname")
+
+  x_data <- df %>%
+    select(-c(colnames(.)[[1]]:ReferenceIntensity)) %>%
+    as.matrix()
+
+  f_data <- df %>%
+    select(c(colnames(.)[[1]]:ReferenceIntensity))
+
+  m <- MSnSet(exprs = x_data, fData = f_data)
+
+  return(m)
+}
+
+
+
+
+utils::globalVariables(
+  c(".", "featureName")
+)
+