diff dimsPredictPuritySingle.R @ 0:e96082a25ff0 draft

"planemo upload for repository https://github.com/computational-metabolomics/mspurity-galaxy commit cb903cd93f9378cfb5eeb68512a54178dcea7bbc-dirty"
author computational-metabolomics
date Wed, 27 Nov 2019 13:42:05 -0500
parents
children 090775983be7
line wrap: on
line diff
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/dimsPredictPuritySingle.R	Wed Nov 27 13:42:05 2019 -0500
@@ -0,0 +1,163 @@
+library(msPurity)
+library(optparse)
+print(sessionInfo())
+
+option_list <- list(
+  make_option(c("--mzML_file"), type="character"),
+  make_option(c("--mzML_files"), type="character"),
+  make_option(c("--mzML_filename"), type="character", default=''),
+  make_option(c("--mzML_galaxy_names"), type="character", default=''),
+  make_option(c("--peaks_file"), type="character"),
+  make_option(c("-o", "--out_dir"), type="character"),
+  make_option("--minoffset", default=0.5),
+  make_option("--maxoffset", default=0.5),
+  make_option("--ilim", default=0.05),
+  make_option("--ppm", default=4),
+  make_option("--dimspy", action="store_true"),
+  make_option("--sim", action="store_true"),
+  make_option("--remove_nas", action="store_true"),
+  make_option("--iwNorm", default="none", type="character"),
+  make_option("--file_num_dimspy", default=1),
+  make_option("--exclude_isotopes", action="store_true"),
+  make_option("--isotope_matrix", type="character")
+)
+
+# store options
+opt<- parse_args(OptionParser(option_list=option_list))
+
+print(sessionInfo())
+print(opt)
+
+print(opt$mzML_files)
+print(opt$mzML_galaxy_names)
+
+str_to_vec <- function(x){
+    print(x)
+    x <- trimws(strsplit(x, ',')[[1]])
+    return(x[x != ""])
+}
+
+find_mzml_file <- function(mzML_files, galaxy_names, mzML_filename){
+    mzML_filename <- trimws(mzML_filename)
+    mzML_files <- str_to_vec(mzML_files)
+    galaxy_names <- str_to_vec(galaxy_names)
+    if (mzML_filename %in% galaxy_names){
+        return(mzML_files[galaxy_names==mzML_filename])
+    }else{
+        stop(paste("mzML file not found - ", mzML_filename))
+    }
+}
+
+
+if (is.null(opt$dimspy)){
+    df <- read.table(opt$peaks_file, header = TRUE, sep='\t')
+    if (file.exists(opt$mzML_file)){
+        mzML_file <- opt$mzML_file
+    }else if (!is.null(opt$mzML_files)){
+        mzML_file <- find_mzml_file(opt$mzML_files, opt$mzML_galaxy_names, 
+                                    opt$mzML_filename)
+    }else{
+        mzML_file <- file.path(opt$mzML_file, filename)    
+    }	
+}else{
+    indf <- read.table(opt$peaks_file,
+                       header = TRUE, sep='\t', stringsAsFactors = FALSE)
+    
+    filename <- colnames(indf)[8:ncol(indf)][opt$file_num_dimspy]
+    print(filename)
+    # check if the data file is mzML or RAW (can only use mzML currently) so
+    # we expect an mzML file of the same name in the same folder
+    indf$i <- indf[,colnames(indf)==filename]
+    indf[,colnames(indf)==filename] <- as.numeric(indf[,colnames(indf)==filename])
+    
+    filename = sub("raw", "mzML", filename, ignore.case = TRUE)
+    print(filename)
+    
+    
+    if (file.exists(opt$mzML_file)){
+        mzML_file <- opt$mzML_file
+    }else if (!is.null(opt$mzML_files)){
+        mzML_file <- find_mzml_file(opt$mzML_files, opt$mzML_galaxy_names, filename)
+    }else{
+        mzML_file <- file.path(opt$mzML_file, filename)    
+    }	
+    
+    # Update the dimspy output with the correct information 
+    df <- indf[4:nrow(indf),]
+    if ('blank_flag' %in% colnames(df)){
+        df <- df[df$blank_flag==1,]
+    }
+    colnames(df)[colnames(df)=='m.z'] <- 'mz'
+    
+    if ('nan' %in% df$mz){
+        df[df$mz=='nan',]$mz <- NA
+    }
+    df$mz <- as.numeric(df$mz)
+}
+
+if (!is.null(opt$remove_nas)){
+  df <- df[!is.na(df$mz),]
+}
+
+if (is.null(opt$isotope_matrix)){
+    im <- NULL
+}else{
+    im <- read.table(opt$isotope_matrix,
+                     header = TRUE, sep='\t', stringsAsFactors = FALSE)
+}
+
+if (is.null(opt$exclude_isotopes)){
+    isotopes <- FALSE
+}else{
+    isotopes <- TRUE
+}
+
+if (is.null(opt$sim)){
+    sim=FALSE
+}else{
+    sim=TRUE
+}
+
+minOffset = as.numeric(opt$minoffset)
+maxOffset = as.numeric(opt$maxoffset)
+
+if (opt$iwNorm=='none'){
+    iwNorm = FALSE
+    iwNormFun = NULL
+}else if (opt$iwNorm=='gauss'){
+    iwNorm = TRUE
+    iwNormFun = msPurity::iwNormGauss(minOff=-minOffset, maxOff=maxOffset)
+}else if (opt$iwNorm=='rcosine'){
+    iwNorm = TRUE
+    iwNormFun = msPurity::iwNormRcosine(minOff=-minOffset, maxOff=maxOffset)
+}else if (opt$iwNorm=='QE5'){
+    iwNorm = TRUE
+    iwNormFun = msPurity::iwNormQE.5()
+}
+
+print('FIRST ROWS OF PEAK FILE')
+print(head(df))
+print(mzML_file)
+predicted <- msPurity::dimsPredictPuritySingle(df$mz,
+                                     filepth=mzML_file,
+                                     minOffset=minOffset,
+                                     maxOffset=maxOffset,
+                                     ppm=opt$ppm,
+                                     mzML=TRUE,
+                                     sim = sim,
+                                     ilim = opt$ilim,
+                                     isotopes = isotopes,
+                                     im = im,
+                                     iwNorm = iwNorm,
+                                     iwNormFun = iwNormFun
+                                     )
+predicted <- cbind(df, predicted)
+
+print(head(predicted))
+print(file.path(opt$out_dir, 'dimsPredictPuritySingle_output.tsv'))
+
+write.table(predicted, 
+            file.path(opt$out_dir, 'dimsPredictPuritySingle_output.tsv'),
+            row.names=FALSE, sep='\t')
+
+