annotate protein_features.R @ 0:b455ec3f4f33 draft

planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
author proteore
date Mon, 12 Nov 2018 11:10:16 -0500
parents
children 2fc914ab92f5
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
1 # Read file and return file content as data.frame
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
2 read_file <- function(path,header){
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
3 file <- try(read.table(path,header=header, sep="\t",stringsAsFactors = FALSE, quote="", check.names = F),silent=TRUE)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
4 if (inherits(file,"try-error")){
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
5 stop("File not found !")
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
6 }else{
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
7 file <- file[!apply(is.na(file) | file == "", 1, all), , drop=FALSE]
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
8 return(file)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
9 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
10 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
11
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
12 order_columns <- function (df,ncol,id_type,file){
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
13 if (id_type=="Uniprot_AC"){ncol=dim.data.frame(file)[2]}
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
14 if (ncol==1){ #already at the right position
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
15 return (df)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
16 } else {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
17 df = df[,c(2:ncol,1,(ncol+1):dim.data.frame(df)[2])]
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
18 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
19 return (df)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
20 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
21
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
22 get_list_from_cp <-function(list){
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
23 list = strsplit(list, "[ \t\n]+")[[1]]
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
24 list = list[list != ""] #remove empty entry
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
25 list = gsub("-.+", "", list) #Remove isoform accession number (e.g. "-2")
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
26 return(list)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
27 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
28
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
29 get_args <- function(){
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
30
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
31 ## Collect arguments
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
32 args <- commandArgs(TRUE)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
33
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
34 ## Default setting when no arguments passed
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
35 if(length(args) < 1) {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
36 args <- c("--help")
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
37 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
38
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
39 ## Help section
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
40 if("--help" %in% args) {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
41 cat("Selection and Annotation HPA
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
42 Arguments:
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
43 --inputtype: type of input (list of id or filename)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
44 --input: input
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
45 --nextprot: path to nextprot information file
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
46 --column: the column number which you would like to apply...
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
47 --header: true/false if your file contains a header
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
48 --type: the type of input IDs (Uniprot_AC/EntrezID)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
49 --pc_features: IsoPoint,SeqLength,MW
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
50 --localization: Chr,SubcellLocations
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
51 --diseases_info: Diseases
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
52 --output: text output filename \n")
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
53
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
54 q(save="no")
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
55 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
56
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
57 parseArgs <- function(x) strsplit(sub("^--", "", x), "=")
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
58 argsDF <- as.data.frame(do.call("rbind", parseArgs(args)))
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
59 args <- as.list(as.character(argsDF$V2))
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
60 names(args) <- argsDF$V1
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
61
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
62 return(args)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
63 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
64
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
65 str2bool <- function(x){
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
66 if (any(is.element(c("t","true"),tolower(x)))){
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
67 return (TRUE)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
68 }else if (any(is.element(c("f","false"),tolower(x)))){
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
69 return (FALSE)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
70 }else{
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
71 return(NULL)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
72 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
73 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
74
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
75 # Get information from neXtProt
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
76 get_nextprot_info <- function(nextprot,input,pc_features,localization,diseases_info){
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
77 if(diseases_info){
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
78 cols = c("NextprotID",pc_features,localization,"Diseases")
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
79 } else {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
80 cols = c("NextprotID",pc_features,localization)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
81 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
82
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
83 cols=cols[cols!="None"]
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
84 info = nextprot[match(input,nextprot$NextprotID),cols]
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
85 return(info)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
86 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
87
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
88 protein_features = function() {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
89
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
90 args <- get_args()
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
91
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
92 #save(args,file="/home/dchristiany/proteore_project/ProteoRE/tools/add_human_protein_features/args.rda")
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
93 #load("/home/dchristiany/proteore_project/ProteoRE/tools/add_human_protein_features/args.rda")
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
94
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
95 #setting variables
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
96 inputtype = args$inputtype
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
97 if (inputtype == "copy_paste") {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
98 input = get_list_from_cp(args$input)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
99 input = input[input!=""]
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
100 } else if (inputtype == "file") {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
101 filename = args$input
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
102 ncol = args$column
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
103 # Check ncol
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
104 if (! as.numeric(gsub("c", "", ncol)) %% 1 == 0) {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
105 stop("Please enter an integer for level")
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
106 } else {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
107 ncol = as.numeric(gsub("c", "", ncol))
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
108 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
109
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
110 header = str2bool(args$header)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
111 file = read_file(filename, header) # Get file content
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
112 input = sapply(file[,ncol],function(x) strsplit(as.character(x),";")[[1]][1],USE.NAMES = F) # Extract Protein IDs list
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
113 if (args$type == "NextprotID" && ! "NextprotID" %in% colnames(file)) { colnames(file)[ncol] <- "NextprotID"
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
114 } else if (args$type == "NextprotID" && "NextprotID" %in% colnames(file) && match("NextprotID",colnames(file))!=ncol ) {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
115 colnames(file)[match("NextprotID",colnames(file))] <- "old_NextprotID"
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
116 colnames(file)[ncol] = "NextprotID"
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
117 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
118 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
119
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
120 # Read reference file
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
121 nextprot = read_file(args$nextprot,T)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
122
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
123 # Parse arguments
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
124 id_type = args$type
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
125 pc_features = strsplit(args$pc_features, ",")[[1]]
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
126 localization = strsplit(args$localization, ",")[[1]]
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
127 diseases_info = str2bool(args$diseases_info)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
128 output = args$output
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
129
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
130 # Change the sample ids if they are Uniprot_AC ids to be able to match them with
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
131 # Nextprot data
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
132 if (id_type=="Uniprot_AC"){
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
133 NextprotID = gsub("^","NX_",input)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
134 if (inputtype == "file" && "NextprotID" %in% colnames(file)){colnames(file)[match("NextprotID",colnames(file))] <- "old_NextprotID"}
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
135 file = cbind(file,NextprotID)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
136 } else if (id_type=="NextprotID") {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
137 if (inputtype == "file") {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
138 NextprotID = file$NextprotID
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
139 } else {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
140 NextprotID = input
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
141 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
142 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
143
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
144 # Select user input protein ids in nextprot
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
145 if ((length(NextprotID[NextprotID %in% nextprot[,1]]))==0){
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
146 write.table("None of the input ids can be found in Nextprot",file=output,sep="\t",quote=FALSE,col.names=TRUE,row.names=FALSE)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
147 } else {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
148 res <- get_nextprot_info(nextprot,NextprotID,pc_features,localization,diseases_info)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
149
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
150 # Write output
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
151 if (inputtype == "copy_paste") {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
152 if (id_type=="Uniprot_AC"){
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
153 output_content = cbind(input, res)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
154 colnames(output_content)[1] = id_type
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
155 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
156 if ("res" %in% colnames(output_content)){colnames(output_content)[which(colnames(output_content)=="res")] = "NexprotID" } #if no features are selected
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
157 } else if (inputtype == "file") {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
158 res = res[!duplicated(res$NextprotID),]
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
159 output_content = merge(file, res,by="NextprotID",incomparables = NA,all.x=T)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
160 output_content = order_columns(output_content,ncol,id_type,file)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
161 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
162 output_content <- as.data.frame(apply(output_content, c(1,2), function(x) gsub("^$|^ $", NA, x))) #convert "" et " " to NA
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
163 write.table(output_content, output, row.names = FALSE, sep = "\t", quote = FALSE)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
164 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
165
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
166 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
167 protein_features()