annotate protein_features.R @ 2:9525c7c2d5d6 draft

planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
author proteore
date Thu, 06 Dec 2018 05:25:41 -0500
parents 2fc914ab92f5
children 7746af0f8209
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){
1
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
13 if (id_type=="Uniprot_AC"){ncol=ncol(file)}
0
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
1
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
75 #take data frame, return data frame
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
76 split_ids_per_line <- function(line,ncol){
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
77
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
78 #print (line)
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
79 header = colnames(line)
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
80 line[ncol] = gsub("[[:blank:]]|\u00A0","",line[ncol])
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
81
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
82 if (length(unlist(strsplit(as.character(line[ncol]),";")))>1) {
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
83 if (length(line)==1 ) {
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
84 lines = as.data.frame(unlist(strsplit(as.character(line[ncol]),";")),stringsAsFactors = F)
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
85 } else {
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
86 if (ncol==1) { #first column
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
87 lines = suppressWarnings(cbind(unlist(strsplit(as.character(line[ncol]),";")), line[2:length(line)]))
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
88 } else if (ncol==length(line)) { #last column
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
89 lines = suppressWarnings(cbind(line[1:ncol-1],unlist(strsplit(as.character(line[ncol]),";"))))
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
90 } else {
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
91 lines = suppressWarnings(cbind(line[1:ncol-1], unlist(strsplit(as.character(line[ncol]),";"),use.names = F), line[(ncol+1):length(line)]))
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
92 }
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
93 }
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
94 colnames(lines)=header
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
95 return(lines)
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
96 } else {
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
97 return(line)
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
98 }
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
99 }
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
100
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
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
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
102 one_id_one_line <-function(tab,ncol){
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
103
2
9525c7c2d5d6 planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents: 1
diff changeset
104 if (ncol(tab)>1){
9525c7c2d5d6 planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents: 1
diff changeset
105
9525c7c2d5d6 planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents: 1
diff changeset
106 tab[,ncol] = sapply(tab[,ncol],function(x) gsub("[[:blank:]]","",x))
9525c7c2d5d6 planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents: 1
diff changeset
107 header=colnames(tab)
9525c7c2d5d6 planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents: 1
diff changeset
108 res=as.data.frame(matrix(ncol=ncol(tab),nrow=0))
9525c7c2d5d6 planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents: 1
diff changeset
109 for (i in 1:nrow(tab) ) {
9525c7c2d5d6 planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents: 1
diff changeset
110 lines = split_ids_per_line(tab[i,],ncol)
9525c7c2d5d6 planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents: 1
diff changeset
111 res = rbind(res,lines)
9525c7c2d5d6 planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents: 1
diff changeset
112 }
9525c7c2d5d6 planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents: 1
diff changeset
113 }else {
9525c7c2d5d6 planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents: 1
diff changeset
114 res = unlist(sapply(tab[,1],function(x) strsplit(x,";")),use.names = F)
9525c7c2d5d6 planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents: 1
diff changeset
115 res = data.frame(res[which(!is.na(res[res!=""]))],stringsAsFactors = F)
9525c7c2d5d6 planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents: 1
diff changeset
116 colnames(res)=colnames(tab)
1
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
117 }
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
118 return(res)
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
119 }
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
120
0
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
121 # Get information from neXtProt
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
122 get_nextprot_info <- function(nextprot,input,pc_features,localization,diseases_info){
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
123 if(diseases_info){
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
124 cols = c("NextprotID",pc_features,localization,"Diseases")
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
125 } else {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
126 cols = c("NextprotID",pc_features,localization)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
127 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
128
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
129 cols=cols[cols!="None"]
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
130 info = nextprot[match(input,nextprot$NextprotID),cols]
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
131 return(info)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
132 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
133
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
134 protein_features = function() {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
135
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
136 args <- get_args()
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
137
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
138 #save(args,file="/home/dchristiany/proteore_project/ProteoRE/tools/add_human_protein_features/args.rda")
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
139 #load("/home/dchristiany/proteore_project/ProteoRE/tools/add_human_protein_features/args.rda")
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
140
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
141 #setting variables
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
142 inputtype = args$inputtype
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
143 if (inputtype == "copy_paste") {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
144 input = get_list_from_cp(args$input)
1
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
145 input = input[which(!is.na(input[input!=""]))]
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
146 ncol=1
0
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
147 } else if (inputtype == "file") {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
148 filename = args$input
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
149 ncol = args$column
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
150 # Check ncol
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
151 if (! as.numeric(gsub("c", "", ncol)) %% 1 == 0) {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
152 stop("Please enter an integer for level")
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
153 } else {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
154 ncol = as.numeric(gsub("c", "", ncol))
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
155 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
156
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
157 header = str2bool(args$header)
1
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
158 file = read_file(filename, header) # Get file content
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
159 if (any(grep(";",file[,ncol]))) {file = one_id_one_line(file,ncol)}
0
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
160 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
161 if (args$type == "NextprotID" && ! "NextprotID" %in% colnames(file)) { colnames(file)[ncol] <- "NextprotID"
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
162 } else if (args$type == "NextprotID" && "NextprotID" %in% colnames(file) && match("NextprotID",colnames(file))!=ncol ) {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
163 colnames(file)[match("NextprotID",colnames(file))] <- "old_NextprotID"
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
164 colnames(file)[ncol] = "NextprotID"
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
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
168 # Read reference file
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
169 nextprot = read_file(args$nextprot,T)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
170
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
171 # Parse arguments
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
172 id_type = args$type
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
173 pc_features = strsplit(args$pc_features, ",")[[1]]
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
174 localization = strsplit(args$localization, ",")[[1]]
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
175 diseases_info = str2bool(args$diseases_info)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
176 output = args$output
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
177
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
178 # 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
179 # Nextprot data
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
180 if (id_type=="Uniprot_AC"){
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
181 NextprotID = gsub("^","NX_",input)
1
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
182 if (inputtype == "file" && "NextprotID" %in% colnames(file)){
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
183 colnames(file)[match("NextprotID",colnames(file))] <- "old_NextprotID"
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
184 file = cbind(file,NextprotID)
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
185 } else {
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
186 file = data.frame(cbind(input,NextprotID))
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
187 colnames(file)[1]="Uniprot-AC"
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
188 }
0
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
189 } else if (id_type=="NextprotID") {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
190 if (inputtype == "file") {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
191 NextprotID = file$NextprotID
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
192 } else {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
193 NextprotID = input
1
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
194 file=data.frame(NextprotID)
0
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
195 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
196 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
197
1
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
198 #Select user input protein ids in nextprot
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
199 NextprotID = unique(NextprotID[which(!is.na(NextprotID[NextprotID!=""]))])
0
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
200 if ((length(NextprotID[NextprotID %in% nextprot[,1]]))==0){
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
201 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
202 } else {
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
203 res <- get_nextprot_info(nextprot,NextprotID,pc_features,localization,diseases_info)
1
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
204 res = res[!duplicated(res$NextprotID),]
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
205 output_content = merge(file, res,by="NextprotID",incomparables = NA,all.x=T)
2fc914ab92f5 planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents: 0
diff changeset
206 output_content = order_columns(output_content,ncol,id_type,file)
0
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
207 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
208 write.table(output_content, output, row.names = FALSE, sep = "\t", quote = FALSE)
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
209 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
210
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
211 }
b455ec3f4f33 planemo upload commit 9760cde192a15cdf3d2dbec05dd867eaa0392bcd-dirty
proteore
parents:
diff changeset
212 protein_features()