annotate add_expression_HPA.R @ 0:234d114cbe3a draft

planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
author proteore
date Mon, 12 Nov 2018 11:05:24 -0500
parents
children 03487ed0f458
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
1 # Read file and return file content as data.frame
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
2 read_file <- function(path,header){
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
3 file <- try(read.csv(path,header=header, sep="\t",stringsAsFactors = FALSE, quote="\"", check.names = F),silent=TRUE)
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
4 if (inherits(file,"try-error")){
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
5 stop("File not found !")
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
6 }else{
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
7 return(file)
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
8 }
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
9 }
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
10
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
11 #convert a string to boolean
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
12 str2bool <- function(x){
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
13 if (any(is.element(c("t","true"),tolower(x)))){
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
14 return (TRUE)
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
15 }else if (any(is.element(c("f","false"),tolower(x)))){
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
16 return (FALSE)
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
17 }else{
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
18 return(NULL)
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
19 }
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
20 }
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
21
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
22 add_expression = function(input, atlas, options) {
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
23 if (all(!input %in% atlas$Ensembl)) {
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
24 return(NULL)
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
25 } else {
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
26 res = atlas[match(input,atlas$Ensembl),c("Ensembl",options)]
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
27 res <- as.data.frame(apply(res, c(1,2), function(x) gsub("^$|^ $", NA, x))) #convert "" et " " to NA
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
28 return(res)
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
29 }
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
30 }
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
31
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
32 main = function() {
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
33 args <- commandArgs(TRUE)
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
34 if(length(args)<1) {
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
35 args <- c("--help")
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
36 }
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
37
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
38 # Help section
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
39 if("--help" %in% args) {
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
40 cat("Selection and Annotation HPA
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
41 Arguments:
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
42 --inputtype: type of input (list of id or filename)
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
43 --input: either a file name (e.g : input.txt) or a list of blank-separated
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
44 ENSG identifiers (e.g : ENSG00000283071 ENSG00000283072)
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
45 --atlas: path to protein atlas file
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
46 --column: the column number which you would like to apply...
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
47 --header: true/false if your file contains a header
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
48 --select: information from HPA to select, maybe:
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
49 RNA.tissue.category,Reliability..IH.,Reliability..IF. (comma-separated)
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
50 --output: text output filename \n")
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
51 q(save="no")
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
52 }
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
53
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
54 # Parse arguments
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
55 parseArgs <- function(x) strsplit(sub("^--", "", x), "=")
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
56 argsDF <- as.data.frame(do.call("rbind", parseArgs(args)))
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
57 args <- as.list(as.character(argsDF$V2))
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
58 names(args) <- argsDF$V1
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
59
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
60 save(args,file="/home/dchristiany/proteore_project/ProteoRE/tools/add_expression_data_HPA/args.rda")
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
61 #load("/home/dchristiany/proteore_project/ProteoRE/tools/add_expression_data_HPA/args.rda")
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
62
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
63 inputtype = args$inputtype
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
64 if (inputtype == "copypaste") {
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
65 input = strsplit(args$input, "[ \t\n]+")[[1]]
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
66 } else if (inputtype == "tabfile") {
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
67 filename = args$input
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
68 ncol = args$column
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
69 # Check ncol
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
70 if (! as.numeric(gsub("c", "", ncol)) %% 1 == 0) {
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
71 stop("Please enter an integer for level")
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
72 } else {
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
73 ncol = as.numeric(gsub("c", "", ncol))
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
74 }
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
75 header = str2bool(args$header)
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
76 file = read_file(filename, header)
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
77 input = unlist(sapply(as.character(file[,ncol]),function(x) rapply(strsplit(x,";"),c),USE.NAMES = FALSE))
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
78 }
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
79
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
80 # Read protein atlas
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
81 protein_atlas = args$atlas
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
82 protein_atlas = read_file(protein_atlas, T)
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
83
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
84 # Add expression
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
85 output = args$output
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
86 options = strsplit(args$select, ",")[[1]]
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
87 res = add_expression(input, protein_atlas, options)
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
88
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
89
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
90 # Write output
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
91 if (is.null(res)) {
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
92 write.table("None of the input ENSG ids are can be found in HPA data file",file=output,sep="\t",quote=FALSE,col.names=TRUE,row.names=FALSE)
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
93 } else {
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
94 if (inputtype == "copypaste") {
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
95 output_content = cbind(as.matrix(input), res)
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
96 colnames(output_content)[1] = "Ensembl"
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
97 } else if (inputtype == "tabfile") {
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
98 output_content = merge(file, res, by.x=ncol, by.y=1, incomparables = NA,all.x=T)
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
99 output_content = output_content[order(output_content[,ncol],decreasing = T),]
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
100 }
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
101 output_content <- as.data.frame(apply(output_content, c(1,2), function(x) gsub("^$|^ $", NA, x)))
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
102 write.table(output_content, output, row.names = FALSE, sep = "\t", quote = FALSE)
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
103 }
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
104 }
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
105
234d114cbe3a planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff changeset
106 main()