Mercurial > repos > proteore > proteore_ms_observation_pepatlas

annotate Get_ms-ms_observations.R @ 10:ef723082fc54 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload commit b061341e1157cccd90e33b945bc17705880b3e57"
author proteore
date Mon, 04 May 2020 03:29:36 -0400
parents 14c890ff105a
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
2
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
1 # Read file and return file content as data.frame
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
2 read_file <- function(path,header){
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
3 file <- try(read.csv(path,header=header, sep="\t",stringsAsFactors = FALSE, quote="\"", check.names = F),silent=TRUE)
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
4 if (inherits(file,"try-error")){
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
5 stop("File not found !")
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
6 }else{
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
7 return(file)
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
8 }
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
9 }
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
10
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
11 str2bool <- function(x){
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
12 if (any(is.element(c("t","true"),tolower(x)))){
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
13 return (TRUE)
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
14 }else if (any(is.element(c("f","false"),tolower(x)))){
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
15 return (FALSE)
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
16 }else{
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
17 return(NULL)
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
18 }
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
19 }
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
20
6
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
21 #take data frame, return data frame
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
22 split_ids_per_line <- function(line,ncol){
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
23
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
24 #print (line)
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
25 header = colnames(line)
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
26 line[ncol] = gsub("[[:blank:]]","",line[ncol])
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
27
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
28 if (length(unlist(strsplit(as.character(line[ncol]),";")))>1) {
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
29 if (length(line)==1 ) {
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
30 lines = as.data.frame(unlist(strsplit(as.character(line[ncol]),";")),stringsAsFactors = F)
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
31 } else {
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
32 if (ncol==1) { #first column
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
33 lines = suppressWarnings(cbind(unlist(strsplit(as.character(line[ncol]),";")), line[2:length(line)]))
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
34 } else if (ncol==length(line)) { #last column
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
35 lines = suppressWarnings(cbind(line[1:ncol-1],unlist(strsplit(as.character(line[ncol]),";"))))
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
36 } else {
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
37 lines = suppressWarnings(cbind(line[1:ncol-1], unlist(strsplit(as.character(line[ncol]),";"),use.names = F), line[(ncol+1):length(line)]))
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
38 }
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
39 }
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
40 colnames(lines)=header
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
41 return(lines)
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
42 } else {
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
43 return(line)
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
44 }
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
45 }
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
46
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
47 #create new lines if there's more than one id per cell in the column in order to have only one id per line
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
48 one_id_one_line <-function(tab,ncol){
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
49 if (ncol(tab)>1){
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
50 tab[,ncol] = sapply(tab[,ncol],function(x) gsub("[[:blank:]]","",x))
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
51 header=colnames(tab)
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
52 res=as.data.frame(matrix(ncol=ncol(tab),nrow=0))
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
53 for (i in 1:nrow(tab) ) {
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
54 lines = split_ids_per_line(tab[i,],ncol)
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
55 res = rbind(res,lines)
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
56 }
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
57 }else {
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
58 res = unlist(sapply(tab[,1],function(x) strsplit(x,";")),use.names = F)
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
59 res = data.frame(res[which(!is.na(res[res!=""]))],stringsAsFactors = F)
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
60 colnames(res)=colnames(tab)
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
61 }
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
62 return(res)
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
63 }
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
64
2
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
65 nb_obs_PeptideAtlas <- function(input, atlas_file) {
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
66 ## Calculate the sum of n_observations for each ID in input
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
67 atlas = read_file(atlas_file, T)
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
68 return(atlas$nb_obs[match(input,atlas$Uniprot_AC)])
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
69 }
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
70
6
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
71 #function to create a list of infos from file path
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
72 extract_info_from_path <- function(path) {
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
73 file_name=strsplit(tail(strsplit(path,"/")[[1]],n=1),"\\.")[[1]][1]
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
74 date=tail(strsplit(file_name,"_")[[1]],n=1)
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
75 tissue=paste(strsplit(file_name,"_")[[1]][1:2],collapse="_")
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
76 return (c(date,tissue,file_name,path))
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
77 }
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
78
7
3e9ea4797b00 planemo upload commit 4ba1ebe7b3f5e3fabf78b5fed7ed0b92e2cbf9e5-dirty
proteore
parents: 6
diff changeset
79 clean_ids <- function(ids){
10
ef723082fc54 "planemo upload commit b061341e1157cccd90e33b945bc17705880b3e57"
proteore
parents: 9
diff changeset
80 ids = gsub(" ",NA,ids)
ef723082fc54 "planemo upload commit b061341e1157cccd90e33b945bc17705880b3e57"
proteore
parents: 9
diff changeset
81 ids = gsub("^$|^ $", NA, ids)
ef723082fc54 "planemo upload commit b061341e1157cccd90e33b945bc17705880b3e57"
proteore
parents: 9
diff changeset
82 ids = gsub("NA",NA,ids)
7
3e9ea4797b00 planemo upload commit 4ba1ebe7b3f5e3fabf78b5fed7ed0b92e2cbf9e5-dirty
proteore
parents: 6
diff changeset
83 return(ids)
3e9ea4797b00 planemo upload commit 4ba1ebe7b3f5e3fabf78b5fed7ed0b92e2cbf9e5-dirty
proteore
parents: 6
diff changeset
84 }
3e9ea4797b00 planemo upload commit 4ba1ebe7b3f5e3fabf78b5fed7ed0b92e2cbf9e5-dirty
proteore
parents: 6
diff changeset
85
2
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
86 main = function() {
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
87 args <- commandArgs(TRUE)
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
88 if(length(args)<1) {
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
89 args <- c("--help")
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
90 }
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
91
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
92 # Help section
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
93 if("--help" %in% args) {
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
94 cat("Selection and Annotation HPA
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
95 Arguments:
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
96 --input_type: type of input (list of id or filename)
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
97 --input: input
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
98 --atlas: list of file(s) path to use
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
99 --output: text output filename \n")
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
100 q(save="no")
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
101 }
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
102
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
103 # Parse arguments
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
104 parseArgs <- function(x) strsplit(sub("^--", "", x), "=")
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
105 argsDF <- as.data.frame(do.call("rbind", parseArgs(args)))
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
106 args <- as.list(as.character(argsDF$V2))
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
107 names(args) <- argsDF$V1
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
108
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
109 # Extract input
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
110 input_type = args$input_type
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
111 if (input_type == "list") {
7
3e9ea4797b00 planemo upload commit 4ba1ebe7b3f5e3fabf78b5fed7ed0b92e2cbf9e5-dirty
proteore
parents: 6
diff changeset
112 input = unlist(strsplit(strsplit(args$input, "[ \t\n]+")[[1]],";"))
2
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
113 } else if (input_type == "file") {
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
114 filename = args$input
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
115 ncol = args$column
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
116 # Check ncol
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
117 if (! as.numeric(gsub("c", "", ncol)) %% 1 == 0) {
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
118 stop("Please enter an integer for level")
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
119 } else {
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
120 ncol = as.numeric(gsub("c", "", ncol))
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
121 }
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
122 header = str2bool(args$header)
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
123 file = read_file(filename, header)
6
e77c0f3e9bab planemo upload commit 22be35397ac49f1368a77fb1515f2293a41c14b3-dirty
proteore
parents: 2
diff changeset
124 file = one_id_one_line(file,ncol) #only one id per line
2
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
125 input = sapply(file[,ncol],function(x) strsplit(as.character(x),";")[[1]][1],USE.NAMES = F)
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
126 }
7
3e9ea4797b00 planemo upload commit 4ba1ebe7b3f5e3fabf78b5fed7ed0b92e2cbf9e5-dirty
proteore
parents: 6
diff changeset
127 input = clean_ids(input)
2
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
128 output = args$output
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
129
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
130 #data_frame building
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
131 paths=strsplit(args$atlas,",")[[1]]
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
132 tmp <- sapply(paths, extract_info_from_path,USE.NAMES = FALSE)
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
133 df <- as.data.frame(t(as.data.frame(tmp)),row.names = c(""),stringsAsFactors = FALSE)
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
134 names(df) <- c("date","tissue","filename","path")
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
135
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
136 # Annotations
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
137 res = sapply(df$path, function(x) nb_obs_PeptideAtlas(input, x), USE.NAMES = FALSE)
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
138 colnames(res)=df$filename
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
139
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
140 # Write output
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
141 if (input_type == "list") {
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
142 res = cbind(as.matrix(input), res)
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
143 colnames(res)[1] = "Uniprot accession number"
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
144 } else if (input_type == "file") {
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
145 res = cbind(file, res)
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
146 }
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
147 res = as.data.frame(apply(res, c(1,2), function(x) gsub("^$|^ $", NA, x)))
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
148 write.table(res, output, row.names = FALSE, sep = "\t", quote = FALSE)
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
149
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
150 }
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
151
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
152 main()
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
153 #Rscript retrieve_peptideatlas.R --input_type="file" --input="test-data/FKW_Lacombe_et_al_2017_OK.txt" --atlas_brain="Human_Brain_201803_PeptideAtlas.txt" --column="c1" --header="true" --output="test-data/PeptideAtlas_output.txt" --atlas_urine="Human_Urine_201803_PeptideAtlas.txt" --atlas="brain,urine"
6ab9d2778f04 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents:
diff changeset
154