Mercurial > repos > lecorguille > mnsbase_readmsdata
annotate msnbase_readmsdata.r @ 0:3f0a218e2ebc draft default tip
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
| author | lecorguille | 
|---|---|
| date | Tue, 03 Apr 2018 11:45:58 -0400 | 
| parents | |
| children | 
| rev | line source | 
|---|---|
| 0 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 1 #!/usr/bin/env Rscript | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 2 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 3 # ----- LOG FILE ----- | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 4 log_file <- file("log.txt", open="wt") | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 5 sink(log_file) | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 6 sink(log_file, type = "output") | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 7 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 8 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 9 # ----- PACKAGE ----- | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 10 cat("\tSESSION INFO\n") | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 11 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 12 #Import the different functions | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 13 source_local <- function(fname){ argv <- commandArgs(trailingOnly=FALSE); base_dir <- dirname(substring(argv[grep("--file=", argv)], 8)); source(paste(base_dir, fname, sep="/")) } | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 14 source_local("lib.r") | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 15 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 16 pkgs <- c("MSnbase","batch") | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 17 loadAndDisplayPackages(pkgs) | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 18 cat("\n\n"); | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 19 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 20 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 21 # ----- ARGUMENTS ----- | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 22 cat("\tARGUMENTS INFO\n") | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 23 args <- parseCommandArgs(evaluate = FALSE) #interpretation of arguments given in command line as an R list of objects | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 24 write.table(as.matrix(args), col.names=F, quote=F, sep='\t') | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 25 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 26 cat("\n\n") | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 27 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 28 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 29 # ----- PROCESSING INFILE ----- | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 30 cat("\tARGUMENTS PROCESSING INFO\n") | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 31 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 32 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 33 cat("\n\n") | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 34 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 35 # ----- INFILE PROCESSING ----- | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 36 cat("\tINFILE PROCESSING INFO\n") | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 37 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 38 # Handle infiles | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 39 if (!exists("singlefile")) singlefile <- NULL | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 40 if (!exists("zipfile")) zipfile <- NULL | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 41 rawFilePath <- getRawfilePathFromArguments(singlefile, zipfile, args) | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 42 zipfile <- rawFilePath$zipfile | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 43 singlefile <- rawFilePath$singlefile | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 44 directory <- retrieveRawfileInTheWorkingDirectory(singlefile, zipfile) | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 45 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 46 # Check some character issues | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 47 md5sumList <- list("origin" = getMd5sum(directory)) | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 48 checkXmlStructure(directory) | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 49 checkFilesCompatibilityWithXcms(directory) | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 50 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 51 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 52 cat("\n\n") | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 53 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 54 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 55 # ----- MAIN PROCESSING INFO ----- | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 56 cat("\tMAIN PROCESSING INFO\n") | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 57 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 58 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 59 cat("\t\tCOMPUTE\n") | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 60 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 61 ## Get the full path to the files | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 62 files <- getMSFiles(directory) | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 63 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 64 cat("\t\t\tCreate a phenodata data.frame\n") | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 65 s_groups <- sapply(files, function(x) tail(unlist(strsplit(dirname(x),"/")), n=1)) | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 66 s_name <- tools::file_path_sans_ext(basename(files)) | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 67 pd <- data.frame(sample_name=s_name, sample_group=s_groups, stringsAsFactors=FALSE) | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 68 print(pd) | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 69 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 70 cat("\t\t\tLoad Raw Data\n") | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 71 raw_data <- readMSData(files=files, pdata = new("NAnnotatedDataFrame", pd), mode="onDisk") | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 72 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 73 # Transform the files absolute pathways into relative pathways | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 74 raw_data@processingData@files <- sub(paste(getwd(), "/", sep="") , "", raw_data@processingData@files) | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 75 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 76 # Create a sampleMetada file | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 77 sampleNamesList <- getSampleMetadata(xdata=raw_data, sampleMetadataOutput="sampleMetadata.tsv") | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 78 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 79 cat("\n\n") | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 80 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 81 # ----- EXPORT ----- | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 82 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 83 cat("\tMSnExp OBJECT INFO\n") | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 84 print(raw_data) | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 85 cat("\t\tphenoData\n") | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 86 print(raw_data@phenoData@data) | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 87 cat("\n\n") | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 88 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 89 #saving R data in .Rdata file to save the variables used in the present tool | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 90 objects2save <- c("raw_data", "zipfile", "singlefile", "md5sumList", "sampleNamesList") | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 91 save(list=objects2save[objects2save %in% ls()], file="readmsdata.RData") | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 92 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 93 | 
| 
3f0a218e2ebc
planemo upload for repository https://github.com/workflow4metabolomics/xcms commit f01148783819c37e474790dbd56619862960448a
 lecorguille parents: diff
changeset | 94 cat("\tDONE\n") | 
