annotate add_protein_features.R @ 18:0a9ae3d7dbf2 draft default tip

"planemo upload commit 7afd4b3ee25f024257ccbac6e51076d25b2a04e7"
author proteore
date Thu, 20 Aug 2020 03:09:52 -0400
parents 7caa90759aba
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
9
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
1 # Read file and return file content as data.frame
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
2 read_file <- function(path,header){
12
1a4cfa7a3a27 planemo upload commit 77279e994f5751c6cd9aa165aa0604db3d241271-dirty
proteore
parents: 11
diff changeset
3 file <- try(read.csv(path,header=header, sep="\t",stringsAsFactors = FALSE, quote="", check.names = F),silent=TRUE)
9
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
4 if (inherits(file,"try-error")){
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
5 stop("File not found !")
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
6 }else{
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
7 file <- file[!apply(is.na(file) | file == "", 1, all), , drop=FALSE]
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
8 return(file)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
9 }
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
10 }
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
11
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
12 order_columns <- function (df,ncol,id_type,file){
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
13 if (id_type=="Uniprot_AC"){ncol=ncol(file)}
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
14 if (ncol==1){ #already at the right position
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
15 return (df)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
16 } else {
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
17 df = df[,c(2:ncol,1,(ncol+1):dim.data.frame(df)[2])]
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
18 }
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
19 return (df)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
20 }
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
21
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
22 get_list_from_cp <-function(list){
16
7caa90759aba planemo upload commit 4ba1ebe7b3f5e3fabf78b5fed7ed0b92e2cbf9e5-dirty
proteore
parents: 13
diff changeset
23 list = gsub(";","\t",list)
9
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
24 list = strsplit(list, "[ \t\n]+")[[1]]
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
25 list = gsub("NA","",list)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
26 list = list[list != ""] #remove empty entry
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
27 list = gsub("-.+", "", list) #Remove isoform accession number (e.g. "-2")
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
28 return(list)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
29 }
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
30
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
31 get_args <- function(){
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
32
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
33 ## Collect arguments
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
34 args <- commandArgs(TRUE)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
35
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
36 ## Default setting when no arguments passed
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
37 if(length(args) < 1) {
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
38 args <- c("--help")
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
39 }
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
40
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
41 ## Help section
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
42 if("--help" %in% args) {
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
43 cat("Selection and Annotation HPA
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
44 Arguments:
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
45 --inputtype: type of input (list of id or filename)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
46 --input: input
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
47 --nextprot: path to nextprot information file
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
48 --column: the column number which you would like to apply...
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
49 --header: true/false if your file contains a header
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
50 --type: the type of input IDs (Uniprot_AC/EntrezID)
18
0a9ae3d7dbf2 "planemo upload commit 7afd4b3ee25f024257ccbac6e51076d25b2a04e7"
proteore
parents: 16
diff changeset
51 --pc_features: IsoPoint,SeqLength,MW,Chr,SubcellLocations,Diseases,protein_name,function,post_trans_mod,protein_family,pathway
9
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
52 --output: text output filename \n")
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
53
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
54 q(save="no")
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
55 }
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
56
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
57 parseArgs <- function(x) strsplit(sub("^--", "", x), "=")
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
58 argsDF <- as.data.frame(do.call("rbind", parseArgs(args)))
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
59 args <- as.list(as.character(argsDF$V2))
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
60 names(args) <- argsDF$V1
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
61
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
62 return(args)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
63 }
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
64
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
65 str2bool <- function(x){
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
66 if (any(is.element(c("t","true"),tolower(x)))){
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
67 return (TRUE)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
68 }else if (any(is.element(c("f","false"),tolower(x)))){
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
69 return (FALSE)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
70 }else{
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
71 return(NULL)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
72 }
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
73 }
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
74
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
75 #take data frame, return data frame
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
76 split_ids_per_line <- function(line,ncol){
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
77
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
78 #print (line)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
79 header = colnames(line)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
80 line[ncol] = gsub("[[:blank:]]|\u00A0","",line[ncol])
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
81
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
82 if (length(unlist(strsplit(as.character(line[ncol]),";")))>1) {
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
83 if (length(line)==1 ) {
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
84 lines = as.data.frame(unlist(strsplit(as.character(line[ncol]),";")),stringsAsFactors = F)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
85 } else {
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
86 if (ncol==1) { #first column
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
87 lines = suppressWarnings(cbind(unlist(strsplit(as.character(line[ncol]),";")), line[2:length(line)]))
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
88 } else if (ncol==length(line)) { #last column
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
89 lines = suppressWarnings(cbind(line[1:ncol-1],unlist(strsplit(as.character(line[ncol]),";"))))
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
90 } else {
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
91 lines = suppressWarnings(cbind(line[1:ncol-1], unlist(strsplit(as.character(line[ncol]),";"),use.names = F), line[(ncol+1):length(line)]))
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
92 }
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
93 }
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
94 colnames(lines)=header
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
95 return(lines)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
96 } else {
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
97 return(line)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
98 }
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
99 }
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
100
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
101 #create new lines if there's more than one id per cell in the columns in order to have only one id per line
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
102 one_id_one_line <-function(tab,ncol){
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
103
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
104 if (ncol(tab)>1){
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
105
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
106 tab[,ncol] = sapply(tab[,ncol],function(x) gsub("[[:blank:]]","",x))
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
107 header=colnames(tab)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
108 res=as.data.frame(matrix(ncol=ncol(tab),nrow=0))
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
109 for (i in 1:nrow(tab) ) {
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
110 lines = split_ids_per_line(tab[i,],ncol)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
111 res = rbind(res,lines)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
112 }
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
113 }else {
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
114 res = unlist(sapply(tab[,1],function(x) strsplit(x,";")),use.names = F)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
115 res = data.frame(res[which(!is.na(res[res!=""]))],stringsAsFactors = F)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
116 colnames(res)=colnames(tab)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
117 }
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
118 return(res)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
119 }
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
120
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
121 # Get information from neXtProt
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
122 get_nextprot_info <- function(nextprot,input,pc_features,localization,diseases_info){
18
0a9ae3d7dbf2 "planemo upload commit 7afd4b3ee25f024257ccbac6e51076d25b2a04e7"
proteore
parents: 16
diff changeset
123 cols = c("NextprotID",pc_features)
9
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
124 cols=cols[cols!="None"]
18
0a9ae3d7dbf2 "planemo upload commit 7afd4b3ee25f024257ccbac6e51076d25b2a04e7"
proteore
parents: 16
diff changeset
125 info = nextprot[match(input,nextprot$NextprotID),intersect(colnames(nextprot),cols)]
9
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
126 return(info)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
127 }
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
128
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
129 protein_features = function() {
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
130
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
131 args <- get_args()
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
132
11
8df559ad14a7 planemo upload commit 27f7b6a4a0139ca47ae874fe2fe6de4bd40ffbcf-dirty
proteore
parents: 9
diff changeset
133 #save(args,file="/home/dchristiany/proteore_project/ProteoRE/tools/add_protein_features/args.rda")
8df559ad14a7 planemo upload commit 27f7b6a4a0139ca47ae874fe2fe6de4bd40ffbcf-dirty
proteore
parents: 9
diff changeset
134 #load("/home/dchristiany/proteore_project/ProteoRE/tools/add_protein_features/args.rda")
9
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
135
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
136 #setting variables
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
137 inputtype = args$inputtype
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
138 if (inputtype == "copy_paste") {
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
139 input = get_list_from_cp(args$input)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
140 file = data.frame(input,stringsAsFactors = F)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
141 ncol=1
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
142 } else if (inputtype == "file") {
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
143 filename = args$input
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
144 ncol = args$column
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
145 # Check ncol
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
146 if (! as.numeric(gsub("c", "", ncol)) %% 1 == 0) {
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
147 stop("Please enter an integer for level")
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
148 } else {
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
149 ncol = as.numeric(gsub("c", "", ncol))
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
150 }
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
151
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
152 header = str2bool(args$header)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
153 file = read_file(filename, header) # Get file content
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
154 if (any(grep(";",file[,ncol]))) {file = one_id_one_line(file,ncol)}
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
155 if (args$type == "NextprotID" && ! "NextprotID" %in% colnames(file)) { colnames(file)[ncol] <- "NextprotID"
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
156 } else if (args$type == "NextprotID" && "NextprotID" %in% colnames(file) && match("NextprotID",colnames(file))!=ncol ) {
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
157 colnames(file)[match("NextprotID",colnames(file))] <- "old_NextprotID"
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
158 colnames(file)[ncol] = "NextprotID"
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
159 }
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
160 }
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
161
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
162 # Read reference file
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
163 nextprot = read_file(args$nextprot,T)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
164
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
165 # Parse arguments
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
166 id_type = args$type
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
167 pc_features = strsplit(args$pc_features, ",")[[1]]
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
168 output = args$output
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
169
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
170 # Change the sample ids if they are Uniprot_AC ids to be able to match them with
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
171 # Nextprot data
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
172 if (id_type=="Uniprot_AC"){
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
173 NextprotID = gsub("^NX_$","",gsub("^","NX_",file[,ncol]))
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
174 file = cbind(file,NextprotID)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
175 if (inputtype=="copy_paste") {colnames(file)[1]="Uniprot-AC"}
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
176 ncol=ncol(file)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
177 }
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
178 NextprotID = file[,ncol]
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
179
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
180 #Select user input protein ids in nextprot
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
181 #NextprotID = unique(NextprotID[which(!is.na(NextprotID[NextprotID!=""]))])
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
182 if (all(!NextprotID %in% nextprot[,1])){
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
183 write.table("None of the input ids can be found in Nextprot",file=output,sep="\t",quote=FALSE,col.names=TRUE,row.names=FALSE)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
184 } else {
18
0a9ae3d7dbf2 "planemo upload commit 7afd4b3ee25f024257ccbac6e51076d25b2a04e7"
proteore
parents: 16
diff changeset
185 res <- get_nextprot_info(nextprot,NextprotID,pc_features)
9
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
186 res = res[!duplicated(res$NextprotID),]
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
187 output_content = merge(file, res,by.x=ncol,by.y="NextprotID",incomparables = NA,all.x=T)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
188 output_content = order_columns(output_content,ncol,id_type,file)
13
0116d444d21f planemo upload commit a957a9b76923f071837aa5d86c1ee2271ebd34fd-dirty
proteore
parents: 12
diff changeset
189 if (id_type=="Uniprot_AC"){output_content = output_content[,-which(colnames(output_content)=="NextprotID")]} #remove nextprotID column
9
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
190 output_content <- as.data.frame(apply(output_content, c(1,2), function(x) gsub("^$|^ $", NA, x))) #convert "" et " " to NA
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
191 write.table(output_content, output, row.names = FALSE, sep = "\t", quote = FALSE)
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
192 }
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
193
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
194 }
0b46a7aead62 planemo upload commit eb7450f36863f02f036cbc52bf5525d68f22bd9e
proteore
parents:
diff changeset
195 protein_features()