Mercurial > repos > lecorguille > xcms_fillpeaks
annotate lib.r @ 6:f00e39a4b0fb draft
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
| author | lecorguille |
|---|---|
| date | Thu, 07 Apr 2016 16:13:02 -0400 |
| parents | |
| children | 8a23c743e6ea |
| rev | line source |
|---|---|
|
6
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
1 # lib.r version="2.0.1" |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
2 #Authors ABiMS TEAM |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
3 #Lib.r for Galaxy Workflow4Metabo |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
4 #version 2.2 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
5 #Based on lib.r 2.1 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
6 #Modifications made by Guitton Yann |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
7 #correct bug in Base Peak Chromatogram (BPC) option, not only TIC when scanrange used in xcmsSet |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
8 #Note if scanrange is used a warning is prompted in R console but do not stop PDF generation |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
9 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
10 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
11 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
12 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
13 #@author Y. Guitton |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
14 getBPC <- function(file,rtcor=NULL, ...) { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
15 object <- xcmsRaw(file) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
16 sel <- profRange(object, ...) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
17 cbind(if (is.null(rtcor)) object@scantime[sel$scanidx] else rtcor ,xcms:::colMax(object@env$profile[sel$massidx,sel$scanidx,drop=FALSE])) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
18 #plotChrom(xcmsRaw(file), base=T) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
19 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
20 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
21 #@author Y. Guitton |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
22 getBPCs <- function (xcmsSet=NULL, pdfname="BPCs.pdf",rt=c("raw","corrected"), scanrange=NULL) { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
23 cat("Creating BIC pdf...\n") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
24 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
25 if (is.null(xcmsSet)) { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
26 cat("Enter an xcmsSet \n") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
27 stop() |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
28 } else { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
29 files <- filepaths(xcmsSet) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
30 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
31 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
32 class<-as.vector(levels(xcmsSet@phenoData[,1])) #sometime phenoData have more than 1 column use first as class |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
33 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
34 classnames<-vector("list",length(class)) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
35 for (i in 1:length(class)){ |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
36 classnames[[i]]<-which( xcmsSet@phenoData[,1]==class[i]) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
37 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
38 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
39 N <- dim(phenoData(xcmsSet))[1] |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
40 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
41 TIC <- vector("list",N) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
42 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
43 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
44 for (j in 1:N) { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
45 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
46 TIC[[j]] <- getBPC(files[j]) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
47 #good for raw |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
48 # seems strange for corrected |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
49 #errors if scanrange used in xcmsSetgeneration |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
50 if (!is.null(xcmsSet) && rt == "corrected") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
51 rtcor <- xcmsSet@rt$corrected[[j]] else |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
52 rtcor <- NULL |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
53 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
54 TIC[[j]] <- getBPC(files[j],rtcor=rtcor) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
55 # TIC[[j]][,1]<-rtcor |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
56 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
57 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
58 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
59 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
60 pdf(pdfname,w=16,h=10) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
61 cols <- rainbow(N) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
62 lty = 1:N |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
63 pch = 1:N |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
64 #search for max x and max y in BPCs |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
65 xlim = range(sapply(TIC, function(x) range(x[,1]))) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
66 ylim = range(sapply(TIC, function(x) range(x[,2]))) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
67 ylim = c(-ylim[2], ylim[2]) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
68 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
69 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
70 ##plot start |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
71 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
72 if (length(class)>2){ |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
73 for (k in 1:(length(class)-1)){ |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
74 for (l in (k+1):length(class)){ |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
75 #print(paste(class[k],"vs",class[l],sep=" ")) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
76 plot(0, 0, type="n", xlim = xlim/60, ylim = ylim, main = paste("Base Peak Chromatograms \n","BPCs_",class[k]," vs ",class[l], sep=""), xlab = "Retention Time (min)", ylab = "BPC") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
77 colvect<-NULL |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
78 for (j in 1:length(classnames[[k]])) { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
79 tic <- TIC[[classnames[[k]][j]]] |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
80 # points(tic[,1]/60, tic[,2], col = cols[i], pch = pch[i], type="l") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
81 points(tic[,1]/60, tic[,2], col = cols[classnames[[k]][j]], pch = pch[classnames[[k]][j]], type="l") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
82 colvect<-append(colvect,cols[classnames[[k]][j]]) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
83 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
84 for (j in 1:length(classnames[[l]])) { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
85 # i=class2names[j] |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
86 tic <- TIC[[classnames[[l]][j]]] |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
87 points(tic[,1]/60, -tic[,2], col = cols[classnames[[l]][j]], pch = pch[classnames[[l]][j]], type="l") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
88 colvect<-append(colvect,cols[classnames[[l]][j]]) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
89 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
90 legend("topright",paste(basename(files[c(classnames[[k]],classnames[[l]])])), col = colvect, lty = lty, pch = pch) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
91 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
92 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
93 }#end if length >2 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
94 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
95 if (length(class)==2){ |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
96 k=1 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
97 l=2 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
98 colvect<-NULL |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
99 plot(0, 0, type="n", xlim = xlim/60, ylim = ylim, main = paste("Base Peak Chromatograms \n","BPCs_",class[k],"vs",class[l], sep=""), xlab = "Retention Time (min)", ylab = "BPC") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
100 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
101 for (j in 1:length(classnames[[k]])) { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
102 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
103 tic <- TIC[[classnames[[k]][j]]] |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
104 # points(tic[,1]/60, tic[,2], col = cols[i], pch = pch[i], type="l") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
105 points(tic[,1]/60, tic[,2], col = cols[classnames[[k]][j]], pch = pch[classnames[[k]][j]], type="l") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
106 colvect<-append(colvect,cols[classnames[[k]][j]]) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
107 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
108 for (j in 1:length(classnames[[l]])) { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
109 # i=class2names[j] |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
110 tic <- TIC[[classnames[[l]][j]]] |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
111 points(tic[,1]/60, -tic[,2], col = cols[classnames[[l]][j]], pch = pch[classnames[[l]][j]], type="l") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
112 colvect<-append(colvect,cols[classnames[[l]][j]]) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
113 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
114 legend("topright",paste(basename(files[c(classnames[[k]],classnames[[l]])])), col = colvect, lty = lty, pch = pch) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
115 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
116 }#end length ==2 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
117 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
118 dev.off() #pdf(pdfname,w=16,h=10) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
119 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
120 invisible(TIC) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
121 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
122 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
123 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
124 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
125 #@author Y. Guitton |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
126 getTIC <- function(file,rtcor=NULL) { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
127 object <- xcmsRaw(file) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
128 cbind(if (is.null(rtcor)) object@scantime else rtcor, rawEIC(object,mzrange=range(object@env$mz))$intensity) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
129 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
130 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
131 ## |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
132 ## overlay TIC from all files in current folder or from xcmsSet, create pdf |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
133 ## |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
134 #@author Y. Guitton |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
135 getTICs <- function(xcmsSet=NULL,files=NULL, pdfname="TICs.pdf",rt=c("raw","corrected")) { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
136 cat("Creating TIC pdf...\n") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
137 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
138 if (is.null(xcmsSet)) { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
139 filepattern <- c("[Cc][Dd][Ff]", "[Nn][Cc]", "([Mm][Zz])?[Xx][Mm][Ll]", "[Mm][Zz][Dd][Aa][Tt][Aa]", "[Mm][Zz][Mm][Ll]") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
140 filepattern <- paste(paste("\\.", filepattern, "$", sep = ""), collapse = "|") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
141 if (is.null(files)) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
142 files <- getwd() |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
143 info <- file.info(files) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
144 listed <- list.files(files[info$isdir], pattern = filepattern, recursive = TRUE, full.names = TRUE) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
145 files <- c(files[!info$isdir], listed) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
146 } else { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
147 files <- filepaths(xcmsSet) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
148 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
149 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
150 class<-as.vector(levels(xcmsSet@phenoData[,1])) #sometime phenoData have more than 1 column use first as class |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
151 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
152 classnames<-vector("list",length(class)) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
153 for (i in 1:length(class)){ |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
154 classnames[[i]]<-which( xcmsSet@phenoData[,1]==class[i]) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
155 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
156 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
157 N <- length(files) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
158 TIC <- vector("list",N) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
159 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
160 for (i in 1:N) { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
161 if (!is.null(xcmsSet) && rt == "corrected") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
162 rtcor <- xcmsSet@rt$corrected[[i]] else |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
163 rtcor <- NULL |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
164 TIC[[i]] <- getTIC(files[i],rtcor=rtcor) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
165 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
166 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
167 pdf(pdfname,w=16,h=10) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
168 cols <- rainbow(N) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
169 lty = 1:N |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
170 pch = 1:N |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
171 #search for max x and max y in TICs |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
172 xlim = range(sapply(TIC, function(x) range(x[,1]))) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
173 ylim = range(sapply(TIC, function(x) range(x[,2]))) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
174 ylim = c(-ylim[2], ylim[2]) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
175 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
176 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
177 ##plot start |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
178 if (length(class)>2){ |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
179 for (k in 1:(length(class)-1)){ |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
180 for (l in (k+1):length(class)){ |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
181 #print(paste(class[k],"vs",class[l],sep=" ")) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
182 plot(0, 0, type="n", xlim = xlim/60, ylim = ylim, main = paste("Total Ion Chromatograms \n","TICs_",class[k]," vs ",class[l], sep=""), xlab = "Retention Time (min)", ylab = "TIC") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
183 colvect<-NULL |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
184 for (j in 1:length(classnames[[k]])) { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
185 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
186 tic <- TIC[[classnames[[k]][j]]] |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
187 # points(tic[,1]/60, tic[,2], col = cols[i], pch = pch[i], type="l") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
188 points(tic[,1]/60, tic[,2], col = cols[classnames[[k]][j]], pch = pch[classnames[[k]][j]], type="l") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
189 colvect<-append(colvect,cols[classnames[[k]][j]]) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
190 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
191 for (j in 1:length(classnames[[l]])) { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
192 # i=class2names[j] |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
193 tic <- TIC[[classnames[[l]][j]]] |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
194 points(tic[,1]/60, -tic[,2], col = cols[classnames[[l]][j]], pch = pch[classnames[[l]][j]], type="l") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
195 colvect<-append(colvect,cols[classnames[[l]][j]]) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
196 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
197 legend("topright",paste(basename(files[c(classnames[[k]],classnames[[l]])])), col = colvect, lty = lty, pch = pch) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
198 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
199 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
200 }#end if length >2 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
201 if (length(class)==2){ |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
202 k=1 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
203 l=2 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
204 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
205 plot(0, 0, type="n", xlim = xlim/60, ylim = ylim, main = paste("Total Ion Chromatograms \n","TICs_",class[k],"vs",class[l], sep=""), xlab = "Retention Time (min)", ylab = "TIC") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
206 colvect<-NULL |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
207 for (j in 1:length(classnames[[k]])) { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
208 tic <- TIC[[classnames[[k]][j]]] |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
209 # points(tic[,1]/60, tic[,2], col = cols[i], pch = pch[i], type="l") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
210 points(tic[,1]/60, tic[,2], col = cols[classnames[[k]][j]], pch = pch[classnames[[k]][j]], type="l") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
211 colvect<-append(colvect,cols[classnames[[k]][j]]) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
212 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
213 for (j in 1:length(classnames[[l]])) { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
214 # i=class2names[j] |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
215 tic <- TIC[[classnames[[l]][j]]] |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
216 points(tic[,1]/60, -tic[,2], col = cols[classnames[[l]][j]], pch = pch[classnames[[l]][j]], type="l") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
217 colvect<-append(colvect,cols[classnames[[l]][j]]) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
218 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
219 legend("topright",paste(basename(files[c(classnames[[k]],classnames[[l]])])), col = colvect, lty = lty, pch = pch) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
220 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
221 }#end length ==2 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
222 dev.off() #pdf(pdfname,w=16,h=10) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
223 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
224 invisible(TIC) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
225 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
226 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
227 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
228 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
229 ## |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
230 ## Get the polarities from all the samples of a condition |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
231 #@author Misharl Monsoor misharl.monsoor@sb-roscoff.fr ABiMS TEAM |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
232 #@author Gildas Le Corguille lecorguille@sb-roscoff.fr ABiMS TEAM |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
233 getSampleMetadata <- function(xcmsSet=NULL, sampleMetadataOutput="sampleMetadata.tsv") { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
234 cat("Creating the sampleMetadata file...\n") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
235 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
236 #Create the sampleMetada dataframe |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
237 sampleMetadata=xset@phenoData |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
238 sampleNamesOrigin=rownames(sampleMetadata) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
239 sampleNamesMakeNames=make.names(sampleNamesOrigin) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
240 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
241 if (any(duplicated(sampleNamesMakeNames))) { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
242 write("\n\nERROR: Usually, R has trouble to deal with special characters in its column names, so it rename them using make.names().\nIn your case, at least two columns after the renaming obtain the same name, thus XCMS will collapse those columns per name.", stderr()) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
243 for (sampleName in sampleNamesOrigin) { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
244 write(paste(sampleName,"\t->\t",make.names(sampleName)),stderr()) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
245 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
246 stop("\n\nERROR: One or more of your files will not be import by xcmsSet. It may due to bad characters in their filenames.") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
247 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
248 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
249 if (!all(sampleNamesOrigin == sampleNamesMakeNames)) { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
250 cat("\n\nWARNING: Usually, R has trouble to deal with special characters in its column names, so it rename them using make.names()\nIn your case, one or more sample names will be renamed in the sampleMetadata and dataMatrix files:\n") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
251 for (sampleName in sampleNamesOrigin) { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
252 cat(paste(sampleName,"\t->\t",make.names(sampleName),"\n")) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
253 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
254 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
255 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
256 sampleMetadata$sampleMetadata=sampleNamesMakeNames |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
257 sampleMetadata=cbind(sampleMetadata["sampleMetadata"],sampleMetadata["class"]) #Reorder columns |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
258 rownames(sampleMetadata)=NULL |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
259 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
260 #Create a list of files name in the current directory |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
261 list_files=xset@filepaths |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
262 #For each sample file, the following actions are done |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
263 for (file in list_files){ |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
264 #Check if the file is in the CDF format |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
265 if (!mzR:::netCDFIsFile(file)){ |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
266 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
267 # If the column isn't exist, with add one filled with NA |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
268 if (is.null(sampleMetadata$polarity)) sampleMetadata$polarity=NA |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
269 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
270 #Create a simple xcmsRaw object for each sample |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
271 xcmsRaw=xcmsRaw(file) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
272 #Extract the polarity (a list of polarities) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
273 polarity=xcmsRaw@polarity |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
274 #Verify if all the scans have the same polarity |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
275 uniq_list=unique(polarity) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
276 if (length(uniq_list)>1){ |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
277 polarity="mixed" |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
278 } else { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
279 polarity=as.character(uniq_list) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
280 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
281 #Transforms the character to obtain only the sample name |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
282 filename=basename(file) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
283 library(tools) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
284 samplename=file_path_sans_ext(filename) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
285 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
286 #Set the polarity attribute |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
287 sampleMetadata$polarity[sampleMetadata$sampleMetadata==samplename]=polarity |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
288 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
289 #Delete xcmsRaw object because it creates a bug for the fillpeaks step |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
290 rm(xcmsRaw) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
291 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
292 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
293 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
294 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
295 write.table(sampleMetadata, sep="\t", quote=FALSE, row.names=FALSE, file=sampleMetadataOutput) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
296 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
297 return(list("sampleNamesOrigin"=sampleNamesOrigin,"sampleNamesMakeNames"=sampleNamesMakeNames)) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
298 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
299 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
300 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
301 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
302 ## |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
303 ## This function check if xcms will found all the files |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
304 ## |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
305 #@author Gildas Le Corguille lecorguille@sb-roscoff.fr ABiMS TEAM |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
306 checkFilesCompatibilityWithXcms <- function(directory) { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
307 cat("Checking files filenames compatibilities with xmcs...\n") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
308 # WHAT XCMS WILL FIND |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
309 filepattern <- c("[Cc][Dd][Ff]", "[Nn][Cc]", "([Mm][Zz])?[Xx][Mm][Ll]","[Mm][Zz][Dd][Aa][Tt][Aa]", "[Mm][Zz][Mm][Ll]") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
310 filepattern <- paste(paste("\\.", filepattern, "$", sep = ""),collapse = "|") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
311 info <- file.info(directory) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
312 listed <- list.files(directory[info$isdir], pattern = filepattern,recursive = TRUE, full.names = TRUE) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
313 files <- c(directory[!info$isdir], listed) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
314 files_abs <- file.path(getwd(), files) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
315 exists <- file.exists(files_abs) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
316 files[exists] <- files_abs[exists] |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
317 files[exists] <- sub("//","/",files[exists]) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
318 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
319 # WHAT IS ON THE FILESYSTEM |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
320 filesystem_filepaths=system(paste("find $PWD/",directory," -not -name '\\.*' -not -path '*conda-env*' -type f -name \"*\"", sep=""), intern=T) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
321 filesystem_filepaths=filesystem_filepaths[grep(filepattern, filesystem_filepaths, perl=T)] |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
322 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
323 # COMPARISON |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
324 if (!is.na(table(filesystem_filepaths %in% files)["FALSE"])) { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
325 write("\n\nERROR: List of the files which will not be imported by xcmsSet",stderr()) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
326 write(filesystem_filepaths[!(filesystem_filepaths %in% files)],stderr()) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
327 stop("\n\nERROR: One or more of your files will not be import by xcmsSet. It may due to bad characters in their filenames.") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
328 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
329 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
330 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
331 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
332 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
333 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
334 ## |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
335 ## This function check if XML contains special caracters. It also checks integrity and completness. |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
336 ## |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
337 #@author Misharl Monsoor misharl.monsoor@sb-roscoff.fr ABiMS TEAM |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
338 checkXmlStructure <- function (directory) { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
339 cat("Checking XML structure...\n") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
340 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
341 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;") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
342 capture=system(cmd,intern=TRUE) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
343 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
344 if (length(capture)>0){ |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
345 #message=paste("The following mzXML or mzML file is incorrect, please check these files first:",capture) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
346 write("\n\nERROR: The following mzXML or mzML file(s) are incorrect, please check these files first:", stderr()) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
347 write(capture, stderr()) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
348 stop("ERROR: xcmsSet cannot continue with incorrect mzXML or mzML files") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
349 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
350 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
351 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
352 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
353 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
354 ## |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
355 ## This function check if XML contain special characters |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
356 ## |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
357 #@author Misharl Monsoor misharl.monsoor@sb-roscoff.fr ABiMS TEAM |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
358 deleteXmlBadCharacters<- function (directory) { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
359 cat("Checking Non ASCII characters in the XML...\n") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
360 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
361 processed=F |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
362 l=system( paste("find",directory, "-not -name '\\.*' -not -path '*conda-env*' -type f -iname '*.*ml*'"),intern=TRUE) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
363 for (i in l){ |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
364 cmd=paste("LC_ALL=C grep '[^ -~]' \"",i,"\"",sep="") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
365 capture=suppressWarnings(system(cmd,intern=TRUE)) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
366 if (length(capture)>0){ |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
367 cmd=paste("perl -i -pe 's/[^[:ascii:]]//g;'",i) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
368 print( paste("WARNING: Non ASCII characters have been removed from the ",i,"file") ) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
369 c=system(cmd,intern=TRUE) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
370 capture="" |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
371 processed=T |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
372 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
373 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
374 if (processed) cat("\n\n") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
375 return(processed) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
376 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
377 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
378 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
379 ## |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
380 ## This function will compute MD5 checksum to check the data integrity |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
381 ## |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
382 #@author Gildas Le Corguille lecorguille@sb-roscoff.fr |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
383 getMd5sum <- function (directory) { |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
384 cat("Compute md5 checksum...\n") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
385 # WHAT XCMS WILL FIND |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
386 filepattern <- c("[Cc][Dd][Ff]", "[Nn][Cc]", "([Mm][Zz])?[Xx][Mm][Ll]","[Mm][Zz][Dd][Aa][Tt][Aa]", "[Mm][Zz][Mm][Ll]") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
387 filepattern <- paste(paste("\\.", filepattern, "$", sep = ""),collapse = "|") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
388 info <- file.info(directory) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
389 listed <- list.files(directory[info$isdir], pattern = filepattern,recursive = TRUE, full.names = TRUE) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
390 files <- c(directory[!info$isdir], listed) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
391 exists <- file.exists(files) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
392 files <- files[exists] |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
393 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
394 library(tools) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
395 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
396 #cat("\n\n") |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
397 |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
398 return(as.matrix(md5sum(files))) |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
399 } |
|
f00e39a4b0fb
planemo upload commit a3229faad6949bbca965d1d7e138cb3c0550780e
lecorguille
parents:
diff
changeset
|
400 |
