annotate add_protein_features.R @ 14:564e620ba3cb draft

planemo upload commit e8aae7c385ce83f4ae7494d53b1c0190e3ef6f9a-dirty
author proteore
date Tue, 14 May 2019 09:43:07 -0400
parents aca66c3b8fbb
children bdbc7620e804
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
4
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
1 # Read file and return file content as data.frame
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
2 read_file <- function(path,header){
5
bb4a5f1b415f planemo upload commit f7974932f7ca782b9a5a8a092f84a0bb0f965e64-dirty
proteore
parents: 4
diff changeset
3 file <- try(read.csv(path,header=header, sep="\t",stringsAsFactors = FALSE, quote="", check.names = F),silent=TRUE)
4
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
4 if (inherits(file,"try-error")){
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
5 stop("File not found !")
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
6 }else{
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
7 file <- file[!apply(is.na(file) | file == "", 1, all), , drop=FALSE]
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
8 return(file)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
9 }
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
10 }
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
11
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
12 order_columns <- function (df,ncol,id_type,file){
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
13 if (id_type=="Uniprot_AC"){ncol=ncol(file)}
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
14 if (ncol==1){ #already at the right position
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
15 return (df)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
16 } else {
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
17 df = df[,c(2:ncol,1,(ncol+1):dim.data.frame(df)[2])]
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
18 }
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
19 return (df)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
20 }
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
21
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
22 get_list_from_cp <-function(list){
14
564e620ba3cb planemo upload commit e8aae7c385ce83f4ae7494d53b1c0190e3ef6f9a-dirty
proteore
parents: 8
diff changeset
23 list = gsub(";","\t",list)
4
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
24 list = strsplit(list, "[ \t\n]+")[[1]]
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
25 list = gsub("NA","",list)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
26 list = list[list != ""] #remove empty entry
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
27 list = gsub("-.+", "", list) #Remove isoform accession number (e.g. "-2")
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
28 return(list)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
29 }
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
30
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
31 get_args <- function(){
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
32
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
33 ## Collect arguments
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
34 args <- commandArgs(TRUE)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
35
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
36 ## Default setting when no arguments passed
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
37 if(length(args) < 1) {
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
38 args <- c("--help")
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
39 }
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
40
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
41 ## Help section
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
42 if("--help" %in% args) {
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
43 cat("Selection and Annotation HPA
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
44 Arguments:
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
45 --inputtype: type of input (list of id or filename)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
46 --input: input
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
47 --nextprot: path to nextprot information file
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
48 --column: the column number which you would like to apply...
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
49 --header: true/false if your file contains a header
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
50 --type: the type of input IDs (Uniprot_AC/EntrezID)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
51 --pc_features: IsoPoint,SeqLength,MW
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
52 --localization: Chr,SubcellLocations
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
53 --diseases_info: Diseases
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
54 --output: text output filename \n")
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
55
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
56 q(save="no")
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
57 }
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
58
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
59 parseArgs <- function(x) strsplit(sub("^--", "", x), "=")
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
60 argsDF <- as.data.frame(do.call("rbind", parseArgs(args)))
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
61 args <- as.list(as.character(argsDF$V2))
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
62 names(args) <- argsDF$V1
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
63
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
64 return(args)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
65 }
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
66
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
67 str2bool <- function(x){
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
68 if (any(is.element(c("t","true"),tolower(x)))){
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
69 return (TRUE)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
70 }else if (any(is.element(c("f","false"),tolower(x)))){
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
71 return (FALSE)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
72 }else{
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
73 return(NULL)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
74 }
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
75 }
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
76
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
77 #take data frame, return data frame
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
78 split_ids_per_line <- function(line,ncol){
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
79
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
80 #print (line)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
81 header = colnames(line)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
82 line[ncol] = gsub("[[:blank:]]|\u00A0","",line[ncol])
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
83
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
84 if (length(unlist(strsplit(as.character(line[ncol]),";")))>1) {
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
85 if (length(line)==1 ) {
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
86 lines = as.data.frame(unlist(strsplit(as.character(line[ncol]),";")),stringsAsFactors = F)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
87 } else {
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
88 if (ncol==1) { #first column
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
89 lines = suppressWarnings(cbind(unlist(strsplit(as.character(line[ncol]),";")), line[2:length(line)]))
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
90 } else if (ncol==length(line)) { #last column
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
91 lines = suppressWarnings(cbind(line[1:ncol-1],unlist(strsplit(as.character(line[ncol]),";"))))
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
92 } else {
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
93 lines = suppressWarnings(cbind(line[1:ncol-1], unlist(strsplit(as.character(line[ncol]),";"),use.names = F), line[(ncol+1):length(line)]))
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
94 }
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
95 }
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
96 colnames(lines)=header
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
97 return(lines)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
98 } else {
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
99 return(line)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
100 }
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
101 }
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
102
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
103 #create new lines if there's more than one id per cell in the columns in order to have only one id per line
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
104 one_id_one_line <-function(tab,ncol){
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
105
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
106 if (ncol(tab)>1){
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
107
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
108 tab[,ncol] = sapply(tab[,ncol],function(x) gsub("[[:blank:]]","",x))
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
109 header=colnames(tab)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
110 res=as.data.frame(matrix(ncol=ncol(tab),nrow=0))
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
111 for (i in 1:nrow(tab) ) {
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
112 lines = split_ids_per_line(tab[i,],ncol)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
113 res = rbind(res,lines)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
114 }
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
115 }else {
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
116 res = unlist(sapply(tab[,1],function(x) strsplit(x,";")),use.names = F)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
117 res = data.frame(res[which(!is.na(res[res!=""]))],stringsAsFactors = F)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
118 colnames(res)=colnames(tab)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
119 }
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
120 return(res)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
121 }
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
122
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
123 # Get information from neXtProt
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
124 get_nextprot_info <- function(nextprot,input,pc_features,localization,diseases_info){
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
125 if(diseases_info){
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
126 cols = c("NextprotID",pc_features,localization,"Diseases")
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
127 } else {
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
128 cols = c("NextprotID",pc_features,localization)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
129 }
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
130
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
131 cols=cols[cols!="None"]
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
132 info = nextprot[match(input,nextprot$NextprotID),cols]
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
133 return(info)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
134 }
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
135
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
136 protein_features = function() {
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
137
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
138 args <- get_args()
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
139
5
bb4a5f1b415f planemo upload commit f7974932f7ca782b9a5a8a092f84a0bb0f965e64-dirty
proteore
parents: 4
diff changeset
140 #save(args,file="/home/dchristiany/proteore_project/ProteoRE/tools/add_protein_features/args.rda")
bb4a5f1b415f planemo upload commit f7974932f7ca782b9a5a8a092f84a0bb0f965e64-dirty
proteore
parents: 4
diff changeset
141 #load("/home/dchristiany/proteore_project/ProteoRE/tools/add_protein_features/args.rda")
4
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
142
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
143 #setting variables
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
144 inputtype = args$inputtype
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
145 if (inputtype == "copy_paste") {
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
146 input = get_list_from_cp(args$input)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
147 file = data.frame(input,stringsAsFactors = F)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
148 ncol=1
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
149 } else if (inputtype == "file") {
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
150 filename = args$input
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
151 ncol = args$column
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
152 # Check ncol
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
153 if (! as.numeric(gsub("c", "", ncol)) %% 1 == 0) {
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
154 stop("Please enter an integer for level")
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
155 } else {
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
156 ncol = as.numeric(gsub("c", "", ncol))
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
157 }
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
158
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
159 header = str2bool(args$header)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
160 file = read_file(filename, header) # Get file content
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
161 if (any(grep(";",file[,ncol]))) {file = one_id_one_line(file,ncol)}
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
162 if (args$type == "NextprotID" && ! "NextprotID" %in% colnames(file)) { colnames(file)[ncol] <- "NextprotID"
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
163 } else if (args$type == "NextprotID" && "NextprotID" %in% colnames(file) && match("NextprotID",colnames(file))!=ncol ) {
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
164 colnames(file)[match("NextprotID",colnames(file))] <- "old_NextprotID"
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
165 colnames(file)[ncol] = "NextprotID"
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
166 }
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
167 }
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
168
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
169 # Read reference file
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
170 nextprot = read_file(args$nextprot,T)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
171
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
172 # Parse arguments
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
173 id_type = args$type
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
174 pc_features = strsplit(args$pc_features, ",")[[1]]
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
175 localization = strsplit(args$localization, ",")[[1]]
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
176 diseases_info = str2bool(args$diseases_info)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
177 output = args$output
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
178
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
179 # Change the sample ids if they are Uniprot_AC ids to be able to match them with
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
180 # Nextprot data
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
181 if (id_type=="Uniprot_AC"){
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
182 NextprotID = gsub("^NX_$","",gsub("^","NX_",file[,ncol]))
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
183 file = cbind(file,NextprotID)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
184 if (inputtype=="copy_paste") {colnames(file)[1]="Uniprot-AC"}
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
185 ncol=ncol(file)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
186 }
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
187 NextprotID = file[,ncol]
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
188
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
189 #Select user input protein ids in nextprot
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
190 #NextprotID = unique(NextprotID[which(!is.na(NextprotID[NextprotID!=""]))])
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
191 if (all(!NextprotID %in% nextprot[,1])){
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
192 write.table("None of the input ids can be found in Nextprot",file=output,sep="\t",quote=FALSE,col.names=TRUE,row.names=FALSE)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
193 } else {
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
194 res <- get_nextprot_info(nextprot,NextprotID,pc_features,localization,diseases_info)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
195 res = res[!duplicated(res$NextprotID),]
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
196 output_content = merge(file, res,by.x=ncol,by.y="NextprotID",incomparables = NA,all.x=T)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
197 output_content = order_columns(output_content,ncol,id_type,file)
8
aca66c3b8fbb planemo upload commit 3e8aad84b7fe9218370364b1aef3ea5779475f9f-dirty
proteore
parents: 5
diff changeset
198 if (id_type=="Uniprot_AC"){output_content = output_content[,-which(colnames(output_content)=="NextprotID")]} #remove nextprotID column
4
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
199 output_content <- as.data.frame(apply(output_content, c(1,2), function(x) gsub("^$|^ $", NA, x))) #convert "" et " " to NA
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
200 write.table(output_content, output, row.names = FALSE, sep = "\t", quote = FALSE)
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
201 }
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
202
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
203 }
759850de6ed2 planemo upload commit c599cfc156c77626df2b674bdfbd437b9f664ab9
proteore
parents:
diff changeset
204 protein_features()