Mercurial > repos > proteore > proteore_expression_rnaseq_abbased
annotate add_expression_HPA.R @ 10:07858c6dbbff draft
"planemo upload commit 330a2527c7e20c4d189a0f1b7a184a8c4610b4b6-dirty"
| author | proteore |
|---|---|
| date | Wed, 22 Jan 2020 15:26:11 +0000 |
| parents | 28d51adf8292 |
| children | f8d3a8a2f2e5 |
| 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){ |
|
10
07858c6dbbff
"planemo upload commit 330a2527c7e20c4d189a0f1b7a184a8c4610b4b6-dirty"
proteore
parents:
9
diff
changeset
|
3 file <- read.csv(path,header=header, sep="\t",stringsAsFactors = FALSE, quote="\"", check.names = F) |
|
07858c6dbbff
"planemo upload commit 330a2527c7e20c4d189a0f1b7a184a8c4610b4b6-dirty"
proteore
parents:
9
diff
changeset
|
4 return(file) |
|
0
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
5 } |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
6 |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
7 #convert a string to boolean |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
8 str2bool <- function(x){ |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
9 if (any(is.element(c("t","true"),tolower(x)))){ |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
10 return (TRUE) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
11 }else if (any(is.element(c("f","false"),tolower(x)))){ |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
12 return (FALSE) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
13 }else{ |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
14 return(NULL) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
15 } |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
16 } |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
17 |
|
6
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
18 stopQuietly <- function(...) { |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
19 blankMsg <- sprintf("\r%s\r", paste(rep(" ", getOption("width")-1L), collapse=" ")); |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
20 stop(simpleError(blankMsg)); |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
21 } # stopQuietly() |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
22 |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
23 check_ensembl_geneids <- function(vector,type) { |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
24 ensembl_geneid_pattern = "^ENS[A-Z]+[0-9]{11}$|^[A-Z]{3}[0-9]{3}[A-Za-z](-[A-Za-z])?$|^CG[0-9]+$|^[A-Z0-9]+[.][0-9]+$|^YM[A-Z][0-9]{3}[a-z][0-9]$" |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
25 res = grepl(ensembl_geneid_pattern,vector) |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
26 if (all(!res)){ |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
27 cat("No Ensembl geneIDs found in entered ids") |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
28 stopQuietly() |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
29 } else if (any(!res)) { |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
30 cat(paste(sep="",collapse = " ",c(sum(!res, na.rm=TRUE),'IDs are not ENSG IDs, please check:\n'))) |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
31 not_geneids <- sapply(vector[which(!res)], function(x) paste(sep="",collapse = "",x,"\n"),USE.NAMES = F) |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
32 cat(not_geneids) |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
33 } |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
34 } |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
35 |
|
0
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
36 add_expression = function(input, atlas, options) { |
|
2
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
37 input <- unique(input[!is.na(input)]) |
|
3
d7d2a4512059
planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents:
2
diff
changeset
|
38 input <- gsub("[[:blank:]]|\u00A0","",input) |
|
0
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
39 if (all(!input %in% atlas$Ensembl)) { |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
40 return(NULL) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
41 } else { |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
42 res = atlas[match(input,atlas$Ensembl),c("Ensembl",options)] |
|
2
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
43 res = res[which(!is.na(res[,1])),] |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
44 row.names(res)=res[,1] |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
45 res=res[2:ncol(res)] |
|
0
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
46 res <- as.data.frame(apply(res, c(1,2), function(x) gsub("^$|^ $", NA, x))) #convert "" et " " to NA |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
47 return(res) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
48 } |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
49 } |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
50 |
|
2
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
51 order_columns <- function (df,ncol){ |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
52 if (ncol==1){ #already at the right position |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
53 return (df) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
54 } else { |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
55 df = df[,c(2:ncol,1,(ncol+1):dim.data.frame(df)[2])] |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
56 } |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
57 return (df) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
58 } |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
59 |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
60 #take data frame, return data frame |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
61 split_ids_per_line <- function(line,ncol){ |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
62 |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
63 #print (line) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
64 header = colnames(line) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
65 line[ncol] = gsub("[[:blank:]]","",line[ncol]) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
66 |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
67 if (length(unlist(strsplit(as.character(line[ncol]),";")))>1) { |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
68 if (length(line)==1 ) { |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
69 lines = as.data.frame(unlist(strsplit(as.character(line[ncol]),";")),stringsAsFactors = F) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
70 } else { |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
71 if (ncol==1) { #first column |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
72 lines = suppressWarnings(cbind(unlist(strsplit(as.character(line[ncol]),";")), line[2:length(line)])) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
73 } else if (ncol==length(line)) { #last column |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
74 lines = suppressWarnings(cbind(line[1:ncol-1],unlist(strsplit(as.character(line[ncol]),";")))) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
75 } else { |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
76 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
|
77 } |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
78 } |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
79 colnames(lines)=header |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
80 return(lines) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
81 } else { |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
82 return(line) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
83 } |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
84 } |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
85 |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
86 #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
|
87 one_id_one_line <-function(tab,ncol){ |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
88 |
|
3
d7d2a4512059
planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents:
2
diff
changeset
|
89 if (ncol(tab)>1){ |
|
d7d2a4512059
planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents:
2
diff
changeset
|
90 |
|
d7d2a4512059
planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents:
2
diff
changeset
|
91 tab[,ncol] = sapply(tab[,ncol],function(x) gsub("[[:blank:]]","",x)) |
|
d7d2a4512059
planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents:
2
diff
changeset
|
92 header=colnames(tab) |
|
d7d2a4512059
planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents:
2
diff
changeset
|
93 res=as.data.frame(matrix(ncol=ncol(tab),nrow=0)) |
|
d7d2a4512059
planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents:
2
diff
changeset
|
94 for (i in 1:nrow(tab) ) { |
|
d7d2a4512059
planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents:
2
diff
changeset
|
95 lines = split_ids_per_line(tab[i,],ncol) |
|
d7d2a4512059
planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents:
2
diff
changeset
|
96 res = rbind(res,lines) |
|
d7d2a4512059
planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents:
2
diff
changeset
|
97 } |
|
d7d2a4512059
planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents:
2
diff
changeset
|
98 }else { |
|
d7d2a4512059
planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents:
2
diff
changeset
|
99 res = unlist(sapply(tab[,1],function(x) strsplit(x,";")),use.names = F) |
|
d7d2a4512059
planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents:
2
diff
changeset
|
100 res = data.frame(res[which(!is.na(res[res!=""]))],stringsAsFactors = F) |
|
d7d2a4512059
planemo upload commit 78ad61e52c2bf8c5ffada89a8eed429a332eb40b-dirty
proteore
parents:
2
diff
changeset
|
101 colnames(res)=colnames(tab) |
|
2
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
102 } |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
103 return(res) |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
104 } |
|
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
105 |
|
6
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
106 get_args <- function(){ |
|
0
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
107 args <- commandArgs(TRUE) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
108 if(length(args)<1) { |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
109 args <- c("--help") |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
110 } |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
111 |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
112 # Help section |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
113 if("--help" %in% args) { |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
114 cat("Selection and Annotation HPA |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
115 Arguments: |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
116 --inputtype: type of input (list of id or filename) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
117 --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
|
118 ENSG identifiers (e.g : ENSG00000283071 ENSG00000283072) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
119 --atlas: path to protein atlas file |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
120 --column: the column number which you would like to apply... |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
121 --header: true/false if your file contains a header |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
122 --select: information from HPA to select, maybe: |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
123 RNA.tissue.category,Reliability..IH.,Reliability..IF. (comma-separated) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
124 --output: text output filename \n") |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
125 q(save="no") |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
126 } |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
127 |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
128 # Parse arguments |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
129 parseArgs <- function(x) strsplit(sub("^--", "", x), "=") |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
130 argsDF <- as.data.frame(do.call("rbind", parseArgs(args))) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
131 args <- as.list(as.character(argsDF$V2)) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
132 names(args) <- argsDF$V1 |
|
6
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
133 |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
134 return(args) |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
135 } |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
136 |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
137 is_col_in_file <- function(file,ncol) { |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
138 is_in_file = (ncol <= ncol(file) && ncol > 0) |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
139 if (!is_in_file){ |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
140 cat(paste(sep = "", collapse = " ", c("Column",ncol,"not found in file") )) |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
141 stopQuietly() |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
142 } |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
143 } |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
144 |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
145 main = function() { |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
146 |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
147 args = get_args() |
|
0
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
148 |
|
1
03487ed0f458
planemo upload commit a0d05bb4c75c1762b1ba63cbb63f8975a87911ed-dirty
proteore
parents:
0
diff
changeset
|
149 #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
|
150 #load("/home/dchristiany/proteore_project/ProteoRE/tools/add_expression_data_HPA/args.rda") |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
151 |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
152 inputtype = args$inputtype |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
153 if (inputtype == "copypaste") { |
|
6
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
154 ids = strsplit(args$input, "[ \t\n]+")[[1]] |
|
0
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
155 } else if (inputtype == "tabfile") { |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
156 filename = args$input |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
157 ncol = args$column |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
158 # Check ncol |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
159 if (! as.numeric(gsub("c", "", ncol)) %% 1 == 0) { |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
160 stop("Please enter an integer for level") |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
161 } else { |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
162 ncol = as.numeric(gsub("c", "", ncol)) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
163 } |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
164 header = str2bool(args$header) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
165 file = read_file(filename, header) |
|
6
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
166 is_col_in_file(file,ncol) |
|
2
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
167 file = one_id_one_line(file,ncol) |
|
6
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
168 ids = unlist(sapply(as.character(file[,ncol]),function(x) rapply(strsplit(x,";"),c),USE.NAMES = FALSE)) |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
169 ids = ids[which(!is.na(ids))] |
|
0
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
170 } |
|
6
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
171 check_ensembl_geneids(ids) |
|
0
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
172 |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
173 # Read protein atlas |
|
10
07858c6dbbff
"planemo upload commit 330a2527c7e20c4d189a0f1b7a184a8c4610b4b6-dirty"
proteore
parents:
9
diff
changeset
|
174 protein_atlas = read_file(args$atlas, T) |
|
0
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
175 |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
176 # Add expression |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
177 output = args$output |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
178 options = strsplit(args$select, ",")[[1]] |
|
6
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
179 res = add_expression(ids, protein_atlas, options) |
|
0
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
180 |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
181 # Write output |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
182 if (is.null(res)) { |
|
6
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
183 write.table("None of the ENSG ids entered can be found in HPA data file",file=output,sep="\t",quote=FALSE,col.names=TRUE,row.names=FALSE) |
|
0
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
184 } else { |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
185 if (inputtype == "copypaste") { |
|
6
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
186 ids <- data.frame(ids) |
|
ea59f5750c51
planemo upload commit fb27a6b5de5cd7b269a41be3c85c593b77aa1b18-dirty
proteore
parents:
3
diff
changeset
|
187 output_content = merge(ids,res,by.x=1,by.y="row.names",incomparables = NA, all.x=T) |
|
0
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
188 colnames(output_content)[1] = "Ensembl" |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
189 } else if (inputtype == "tabfile") { |
|
2
70e3b70ded22
planemo upload commit f4517998e11df15ce84f98d2b209e5c43b572c8d-dirty
proteore
parents:
1
diff
changeset
|
190 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
|
191 output_content = order_columns(output_content,ncol) |
|
0
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
192 } |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
193 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
|
194 write.table(output_content, output, row.names = FALSE, sep = "\t", quote = FALSE) |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
195 } |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
196 } |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
197 |
|
234d114cbe3a
planemo upload commit 2ee714c3e3d1cce461125dbc041904e03ea8bac1-dirty
proteore
parents:
diff
changeset
|
198 main() |
