Mercurial > repos > lecorguille > ipo
annotate lib.r @ 4:9fa5856f6184 draft
planemo upload commit 98cb24e57c0ce8ac3294e8c7e87aee337ac29f62
author | lecorguille |
---|---|
date | Thu, 28 Jul 2016 11:11:38 -0400 |
parents | 9f1e18bc8ce3 |
children | 03fdfbd914ab |
rev | line source |
---|---|
3
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
1 ## |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
2 ## This function launch IPO functions to get the best parameters for xcmsSet |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
3 ## A sample among the whole dataset is used to save time |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
4 ## |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
5 ipo4xcmsSet = function(directory, parametersOutput, listArguments, samplebyclass=4) { |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
6 setwd(directory) |
4
9fa5856f6184
planemo upload commit 98cb24e57c0ce8ac3294e8c7e87aee337ac29f62
lecorguille
parents:
3
diff
changeset
|
7 |
3
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
8 files = list.files(".", recursive=T) # "KO/ko15.CDF" "KO/ko16.CDF" "WT/wt15.CDF" "WT/wt16.CDF" |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
9 files_classes = basename(dirname(files)) # "KO", "KO", "WT", "WT" |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
10 |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
11 mzmlfile = files |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
12 if (samplebyclass > 0) { |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
13 #random selection of N files for IPO in each class |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
14 classes<-unique(basename(dirname(files))) |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
15 mzmlfile = NULL |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
16 for (class_i in classes){ |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
17 files_class_i = files[files_classes==class_i] |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
18 if (samplebyclass > length(files_class_i)) { |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
19 mzmlfile = c(mzmlfile, files_class_i) |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
20 } else { |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
21 mzmlfile = c(mzmlfile,sample(files_class_i,samplebyclass)) |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
22 } |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
23 } |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
24 } |
4
9fa5856f6184
planemo upload commit 98cb24e57c0ce8ac3294e8c7e87aee337ac29f62
lecorguille
parents:
3
diff
changeset
|
25 #@TODO: else, must we keep the RData to been use directly by group? |
3
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
26 |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
27 cat("\t\tSamples used:\n") |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
28 print(mzmlfile) |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
29 |
4
9fa5856f6184
planemo upload commit 98cb24e57c0ce8ac3294e8c7e87aee337ac29f62
lecorguille
parents:
3
diff
changeset
|
30 peakpickingParameters = getDefaultXcmsSetStartingParams(listArguments[["method"]]) #get default parameters of IPO |
9fa5856f6184
planemo upload commit 98cb24e57c0ce8ac3294e8c7e87aee337ac29f62
lecorguille
parents:
3
diff
changeset
|
31 |
9fa5856f6184
planemo upload commit 98cb24e57c0ce8ac3294e8c7e87aee337ac29f62
lecorguille
parents:
3
diff
changeset
|
32 # filter listArguments to only get releavant parameters and complete with those that are not declared |
9fa5856f6184
planemo upload commit 98cb24e57c0ce8ac3294e8c7e87aee337ac29f62
lecorguille
parents:
3
diff
changeset
|
33 peakpickingParametersUser = c(listArguments[names(listArguments) %in% names(peakpickingParameters)], peakpickingParameters[!(names(peakpickingParameters) %in% names(listArguments))]) |
9fa5856f6184
planemo upload commit 98cb24e57c0ce8ac3294e8c7e87aee337ac29f62
lecorguille
parents:
3
diff
changeset
|
34 peakpickingParametersUser$verbose.columns = TRUE |
3
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
35 |
4
9fa5856f6184
planemo upload commit 98cb24e57c0ce8ac3294e8c7e87aee337ac29f62
lecorguille
parents:
3
diff
changeset
|
36 #peakpickingParametersUser$profparam <- list(step=0.005) #not yet used by IPO have to think of it for futur improvement |
9fa5856f6184
planemo upload commit 98cb24e57c0ce8ac3294e8c7e87aee337ac29f62
lecorguille
parents:
3
diff
changeset
|
37 resultPeakpicking = optimizeXcmsSet(mzmlfile, peakpickingParametersUser, subdir="IPO_results") #some images generated by IPO |
3
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
38 |
4
9fa5856f6184
planemo upload commit 98cb24e57c0ce8ac3294e8c7e87aee337ac29f62
lecorguille
parents:
3
diff
changeset
|
39 # export |
9fa5856f6184
planemo upload commit 98cb24e57c0ce8ac3294e8c7e87aee337ac29f62
lecorguille
parents:
3
diff
changeset
|
40 resultPeakpicking_best_settings_parameters = resultPeakpicking$best_settings$parameters[!(names(resultPeakpicking$best_settings$parameters) %in% c("nSlaves","verbose.columns"))] |
9fa5856f6184
planemo upload commit 98cb24e57c0ce8ac3294e8c7e87aee337ac29f62
lecorguille
parents:
3
diff
changeset
|
41 write.table(resultPeakpicking_best_settings_parameters, file=parametersOutput, sep="\t",row.names=FALSE) #can be read by user |
3
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
42 |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
43 } |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
44 |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
45 |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
46 |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
47 |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
48 ## |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
49 ## This function check if xcms will found all the files |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
50 ## |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
51 #@author Gildas Le Corguille lecorguille@sb-roscoff.fr ABiMS TEAM |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
52 checkFilesCompatibilityWithXcms <- function(directory) { |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
53 cat("Checking files filenames compatibilities with xmcs...\n") |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
54 # WHAT XCMS WILL FIND |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
55 filepattern <- c("[Cc][Dd][Ff]", "[Nn][Cc]", "([Mm][Zz])?[Xx][Mm][Ll]","[Mm][Zz][Dd][Aa][Tt][Aa]", "[Mm][Zz][Mm][Ll]") |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
56 filepattern <- paste(paste("\\.", filepattern, "$", sep = ""),collapse = "|") |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
57 info <- file.info(directory) |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
58 listed <- list.files(directory[info$isdir], pattern = filepattern,recursive = TRUE, full.names = TRUE) |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
59 files <- c(directory[!info$isdir], listed) |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
60 files_abs <- file.path(getwd(), files) |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
61 exists <- file.exists(files_abs) |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
62 files[exists] <- files_abs[exists] |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
63 files[exists] <- sub("//","/",files[exists]) |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
64 |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
65 # WHAT IS ON THE FILESYSTEM |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
66 filesystem_filepaths=system(paste("find $PWD/",directory," -not -name '\\.*' -not -path '*conda-env*' -type f -name \"*\"", sep=""), intern=T) |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
67 filesystem_filepaths=filesystem_filepaths[grep(filepattern, filesystem_filepaths, perl=T)] |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
68 |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
69 # COMPARISON |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
70 if (!is.na(table(filesystem_filepaths %in% files)["FALSE"])) { |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
71 write("\n\nERROR: List of the files which will not be imported by xcmsSet",stderr()) |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
72 write(filesystem_filepaths[!(filesystem_filepaths %in% files)],stderr()) |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
73 stop("\n\nERROR: One or more of your files will not be import by xcmsSet. It may due to bad characters in their filenames.") |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
74 |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
75 } |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
76 } |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
77 |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
78 |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
79 |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
80 ## |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
81 ## This function check if XML contains special caracters. It also checks integrity and completness. |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
82 ## |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
83 #@author Misharl Monsoor misharl.monsoor@sb-roscoff.fr ABiMS TEAM |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
84 checkXmlStructure <- function (directory) { |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
85 cat("Checking XML structure...\n") |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
86 |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
87 cmd=paste("IFS=$'\n'; for xml in $(find",directory,"-not -name '\\.*' -not -path '*conda-env*' -type f -iname '*.*ml*'); do if [ $(xmllint --nonet --noout \"$xml\" 2> /dev/null; echo $?) -gt 0 ]; then echo $xml;fi; done;") |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
88 capture=system(cmd,intern=TRUE) |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
89 |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
90 if (length(capture)>0){ |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
91 #message=paste("The following mzXML or mzML file is incorrect, please check these files first:",capture) |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
92 write("\n\nERROR: The following mzXML or mzML file(s) are incorrect, please check these files first:", stderr()) |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
93 write(capture, stderr()) |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
94 stop("ERROR: xcmsSet cannot continue with incorrect mzXML or mzML files") |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
95 } |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
96 |
9f1e18bc8ce3
planemo upload commit ceb25d29a013b58d3476323f202276e7c876648a-dirty
lecorguille
parents:
diff
changeset
|
97 } |