Mercurial > repos > proteore > proteore_expression_rnaseq_abbased
annotate add_expression_HPA.R @ 2:70e3b70ded22 draft
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
| author | proteore |
|---|---|
| date | Tue, 04 Dec 2018 05:47:23 -0500 |
| parents | 03487ed0f458 |
| children | d7d2a4512059 |
| 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) { |
|
2
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
23 input <- unique(input[!is.na(input)]) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
24 input <- gsub("[[:blank:]]","",input) |
|
0
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
25 if (all(!input %in% atlas$Ensembl)) { |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
26 return(NULL) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
27 } else { |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
28 res = atlas[match(input,atlas$Ensembl),c("Ensembl",options)] |
|
2
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
29 res = res[which(!is.na(res[,1])),] |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
30 row.names(res)=res[,1] |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
31 res=res[2:ncol(res)] |
|
0
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
32 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
|
33 return(res) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
34 } |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
35 } |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
36 |
|
2
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
37 order_columns <- function (df,ncol){ |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
38 if (ncol==1){ #already at the right position |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
39 return (df) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
40 } else { |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
41 df = df[,c(2:ncol,1,(ncol+1):dim.data.frame(df)[2])] |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
42 } |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
43 return (df) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
44 } |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
45 |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
46 #take data frame, return data frame |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
47 split_ids_per_line <- function(line,ncol){ |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
48 |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
49 #print (line) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
50 header = colnames(line) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
51 line[ncol] = gsub("[[:blank:]]","",line[ncol]) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
52 |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
53 if (length(unlist(strsplit(as.character(line[ncol]),";")))>1) { |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
54 if (length(line)==1 ) { |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
55 lines = as.data.frame(unlist(strsplit(as.character(line[ncol]),";")),stringsAsFactors = F) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
56 } else { |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
57 if (ncol==1) { #first column |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
58 lines = suppressWarnings(cbind(unlist(strsplit(as.character(line[ncol]),";")), line[2:length(line)])) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
59 } else if (ncol==length(line)) { #last column |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
60 lines = suppressWarnings(cbind(line[1:ncol-1],unlist(strsplit(as.character(line[ncol]),";")))) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
61 } else { |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
62 lines = suppressWarnings(cbind(line[1:ncol-1], unlist(strsplit(as.character(line[ncol]),";"),use.names = F), line[(ncol+1):length(line)])) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
63 } |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
64 } |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
65 colnames(lines)=header |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
66 return(lines) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
67 } else { |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
68 return(line) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
69 } |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
70 } |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
71 |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
72 #create new lines if there's more than one id per cell in the columns in order to have only one id per line |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
73 one_id_one_line <-function(tab,ncol){ |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
74 |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
75 tab[,ncol] = sapply(tab[,ncol],function(x) gsub("[[:blank:]]","",x)) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
76 header=colnames(tab) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
77 res=as.data.frame(matrix(ncol=ncol(tab),nrow=0)) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
78 for (i in 1:nrow(tab) ) { |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
79 lines = split_ids_per_line(tab[i,],ncol) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
80 res = rbind(res,lines) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
81 } |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
82 return(res) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
83 } |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
84 |
|
0
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
85 main = function() { |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
86 args <- commandArgs(TRUE) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
87 if(length(args)<1) { |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
88 args <- c("--help") |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
89 } |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
90 |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
91 # Help section |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
92 if("--help" %in% args) { |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
93 cat("Selection and Annotation HPA |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
94 Arguments: |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
95 --inputtype: type of input (list of id or filename) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
96 --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
|
97 ENSG identifiers (e.g : ENSG00000283071 ENSG00000283072) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
98 --atlas: path to protein atlas file |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
99 --column: the column number which you would like to apply... |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
100 --header: true/false if your file contains a header |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
101 --select: information from HPA to select, maybe: |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
102 RNA.tissue.category,Reliability..IH.,Reliability..IF. (comma-separated) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
103 --output: text output filename \n") |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
104 q(save="no") |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
105 } |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
106 |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
107 # Parse arguments |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
108 parseArgs <- function(x) strsplit(sub("^--", "", x), "=") |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
109 argsDF <- as.data.frame(do.call("rbind", parseArgs(args))) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
110 args <- as.list(as.character(argsDF$V2)) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
111 names(args) <- argsDF$V1 |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
112 |
|
1
03487ed0f458
planemo upload commit a0d05bb4c75c1762b1ba63cbb63f8975a87911ed-dirty
proteore
parents:
0
diff
changeset
|
113 #save(args,file="/home/dchristiany/proteore_project/ProteoRE/tools/add_expression_data_HPA/args.rda") |
|
0
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
114 #load("/home/dchristiany/proteore_project/ProteoRE/tools/add_expression_data_HPA/args.rda") |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
115 |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
116 inputtype = args$inputtype |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
117 if (inputtype == "copypaste") { |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
118 input = strsplit(args$input, "[ \t\n]+")[[1]] |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
119 } else if (inputtype == "tabfile") { |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
120 filename = args$input |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
121 ncol = args$column |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
122 # Check ncol |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
123 if (! as.numeric(gsub("c", "", ncol)) %% 1 == 0) { |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
124 stop("Please enter an integer for level") |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
125 } else { |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
126 ncol = as.numeric(gsub("c", "", ncol)) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
127 } |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
128 header = str2bool(args$header) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
129 file = read_file(filename, header) |
|
2
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
130 file = one_id_one_line(file,ncol) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
131 input = unlist(sapply(as.character(file[,ncol]),function(x) rapply(strsplit(x,";"),c),USE.NAMES = FALSE)) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
132 input = input[which(!is.na(input))] |
|
0
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
133 } |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
134 |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
135 # Read protein atlas |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
136 protein_atlas = args$atlas |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
137 protein_atlas = read_file(protein_atlas, T) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
138 |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
139 # Add expression |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
140 output = args$output |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
141 options = strsplit(args$select, ",")[[1]] |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
142 res = add_expression(input, protein_atlas, options) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
143 |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
144 # Write output |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
145 if (is.null(res)) { |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
146 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
|
147 } else { |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
148 if (inputtype == "copypaste") { |
|
2
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
149 input <- data.frame(input) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
150 output_content = merge(input,res,by.x=1,by.y="row.names",incomparables = NA, all.x=T) |
|
0
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
151 colnames(output_content)[1] = "Ensembl" |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
152 } else if (inputtype == "tabfile") { |
|
2
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
153 output_content = merge(file, res, by.x=ncol, by.y="row.names", incomparables = NA, all.x=T) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
154 output_content = order_columns(output_content,ncol) |
|
0
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
155 } |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
156 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
|
157 write.table(output_content, output, row.names = FALSE, sep = "\t", quote = FALSE) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
158 } |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
159 } |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
160 |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
161 main() |
