Mercurial > repos > proteore > proteore_prot_features_mouse

annotate add_protein_features_mouse.R @ 0:b9f57cba3172 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
author proteore
date Tue, 14 May 2019 09:47:13 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
1 options(warn=-1) #TURN OFF WARNINGS !!!!!!
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
2
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
3 # Read file and return file content as data.frame
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
4 order_columns <- function (df,ncol,file){
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
5 if (ncol==1){ #already at the right position
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
6 return (df)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
7 } else {
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
8 df = df[,c(2:ncol,1,(ncol+1):dim.data.frame(df)[2])]
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
9 }
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
10 return (df)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
11 }
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
12
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
13 get_list_from_cp <-function(list){
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
14 list = gsub(";","\t",list)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
15 list = strsplit(list, "[ \t\n]+")[[1]]
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
16 list = gsub("NA","",list)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
17 list = list[list != ""] #remove empty entry
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
18 list = gsub("-.+", "", list) #Remove isoform accession number (e.g. "-2")
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
19 return(list)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
20 }
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
21
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
22 check_ids <- function(vector,type) {
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
23 uniprot_pattern = "^([OPQ][0-9][A-Z0-9]{3}[0-9]|[A-NR-Z][0-9]([A-Z][A-Z0-9]{2}[0-9]){1,2})$"
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
24 entrez_id = "^([0-9]+|[A-Z]{1,2}_[0-9]+|[A-Z]{1,2}_[A-Z]{1,4}[0-9]+)$"
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
25 if (type == "entrez")
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
26 return(grepl(entrez_id,vector))
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
27 else if (type == "uniprot") {
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
28 return(grepl(uniprot_pattern,vector))
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
29 }
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
30 }
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
31
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
32 get_args <- function(){
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
33
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
34 ## Collect arguments
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
35 args <- commandArgs(TRUE)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
36
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
37 ## Default setting when no arguments passed
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
38 if(length(args) < 1) {
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
39 args <- c("--help")
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
40 }
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
41
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
42 ## Help section
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
43 if("--help" %in% args) {
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
44 cat("Selection and Annotation HPA
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
45 Arguments:
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
46 --inputtype: type of input (list of id or filename)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
47 --input: input
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
48 --uniprot_file: path to uniprot reference file
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
49 --column: the column number which you would like to apply...
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
50 --header: true/false if your file contains a header
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
51 --pc_features: IsoPoint,SeqLength,MW
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
52 --output: text output filename \n")
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
53
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
54 q(save="no")
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
55 }
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
56
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
57 parseArgs <- function(x) strsplit(sub("^--", "", x), "=")
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
58 argsDF <- as.data.frame(do.call("rbind", parseArgs(args)))
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
59 args <- as.list(as.character(argsDF$V2))
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
60 names(args) <- argsDF$V1
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
61
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
62 return(args)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
63 }
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
64
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
65 str2bool <- function(x){
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
66 if (any(is.element(c("t","true"),tolower(x)))){
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
67 return (TRUE)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
68 }else if (any(is.element(c("f","false"),tolower(x)))){
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
69 return (FALSE)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
70 }else{
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
71 return(NULL)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
72 }
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
73 }
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
74
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
75 #take data frame, return data frame
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
76 split_ids_per_line <- function(line,ncol){
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
77
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
78 #print (line)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
79 header = colnames(line)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
80 line[ncol] = gsub("[[:blank:]]|\u00A0","",line[ncol])
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
81
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
82 if (length(unlist(strsplit(as.character(line[ncol]),";")))>1) {
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
83 if (length(line)==1 ) {
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
84 lines = as.data.frame(unlist(strsplit(as.character(line[ncol]),";")),stringsAsFactors = F)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
85 } else {
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
86 if (ncol==1) { #first column
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
87 lines = suppressWarnings(cbind(unlist(strsplit(as.character(line[ncol]),";")), line[2:length(line)]))
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
88 } else if (ncol==length(line)) { #last column
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
89 lines = suppressWarnings(cbind(line[1:ncol-1],unlist(strsplit(as.character(line[ncol]),";"))))
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
90 } else {
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
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)]))
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
92 }
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
93 }
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
94 colnames(lines)=header
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
95 return(lines)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
96 } else {
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
97 return(line)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
98 }
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
99 }
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
100
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
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
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
102 one_id_one_line <-function(tab,ncol){
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
103
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
104 if (ncol(tab)>1){
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
105
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
106 tab[,ncol] = sapply(tab[,ncol],function(x) gsub("[[:blank:]]","",x))
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
107 header=colnames(tab)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
108 res=as.data.frame(matrix(ncol=ncol(tab),nrow=0))
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
109 for (i in 1:nrow(tab) ) {
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
110 lines = split_ids_per_line(tab[i,],ncol)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
111 res = rbind(res,lines)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
112 }
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
113 }else {
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
114 res = unlist(sapply(tab[,1],function(x) strsplit(x,";")),use.names = F)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
115 res = data.frame(res[which(!is.na(res[res!=""]))],stringsAsFactors = F)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
116 colnames(res)=colnames(tab)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
117 }
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
118 return(res)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
119 }
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
120
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
121 # Get information from neXtProt
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
122 get_uniprot_info <- function(ids,pc_features,uniprot_file){
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
123
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
124 cols = c("Entry",pc_features)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
125 cols=cols[cols!="None"]
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
126 info = uniprot_file[match(ids,uniprot_file$Entry),cols]
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
127 colnames(info)[1]="UniProt-AC"
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
128
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
129 return(info)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
130 }
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
131
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
132 main <- function() {
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
133
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
134 args <- get_args()
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
135
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
136 #save(args,file="/home/dchristiany/proteore_project/ProteoRE/tools/add_protein_features_mouse/args.rda")
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
137 #load("/home/dchristiany/proteore_project/ProteoRE/tools/add_protein_features_mouse/args.rda")
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
138
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
139 #setting variables
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
140 inputtype = args$inputtype
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
141 if (inputtype == "copy_paste") {
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
142 ids = get_list_from_cp(args$input)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
143
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
144 #Check for UniProt-AC ids
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
145 if (all(!check_ids(ids,'uniprot'))){
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
146 stop ("No UniProt-AC found in ids.")
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
147 } else if (any(!check_ids(ids,'uniprot'))) {
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
148 print ('Some ids in ids are not uniprot-AC:')
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
149 print (ids[which(!check_ids(ids,'uniprot'))])
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
150 }
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
151 file = data.frame(ids,stringsAsFactors = F)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
152 ncol=1
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
153
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
154 } else if (inputtype == "file") {
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
155 filename = args$input
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
156 ncol = args$column
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
157 # Check ncol
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
158 if (! as.numeric(gsub("c", "", ncol)) %% 1 == 0) {
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
159 stop("Please enter an integer for level")
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
160 } else {
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
161 ncol = as.numeric(gsub("c", "", ncol))
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
162 }
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
163
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
164 header = str2bool(args$header)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
165 file = read.table(filename, header = header , sep = "\t", fill = TRUE, stringsAsFactors = FALSE, quote="", check.names = F, comment.char = "") # Get file content
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
166 if (any(grep(";",file[,ncol]))) {file = one_id_one_line(file,ncol)}
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
167 ids=file[,ncol]
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
168 }
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
169
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
170 # Read reference file
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
171 uniprot_file <- read.table(args$uniprot_file, header = TRUE , sep = "\t", fill = TRUE,stringsAsFactors = FALSE, quote="", check.names = F, comment.char = "")
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
172
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
173 # Parse arguments
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
174 pc_features = gsub("__ob__","[",gsub("__cb__","]",strsplit(args$pc_features, ",")[[1]]))
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
175 output = args$output
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
176
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
177 #output file
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
178 res <- get_uniprot_info(ids,pc_features,uniprot_file)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
179 res = res[!duplicated(res$`UniProt-AC`),]
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
180 output_content = merge(file, res,by.x=ncol,by.y="UniProt-AC",incomparables = NA,all.x=T)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
181 output_content = order_columns(output_content,ncol,file)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
182 output_content <- as.data.frame(apply(output_content, c(1,2), function(x) gsub("^$|^ $", NA, x))) #convert "" et " " to NA
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
183 write.table(output_content, output, row.names = FALSE, sep = "\t", quote = FALSE)
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
184
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
185 }
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
186
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
187 if(!interactive()) {
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
188 main()
b9f57cba3172 planemo upload commit e507ac5916a604c5fd7fd2af8074cc2fbb6707fe-dirty
proteore
parents:
diff changeset
189 }