annotate PathView.R @ 11:6d5c0ff2b0bd draft

planemo upload commit 3cfb63ac3585431e7cdcc0a0bd11c33b946a6751-dirty
author proteore
date Fri, 14 Sep 2018 09:52:28 -0400
parents e8d6e50f5d38
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
1 #!/usr/bin/Rscript
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
2 #Rscript made for mapping genesID on KEGG pathway with Pathview package
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
3 #input : csv file containing ids (uniprot or geneID) to map, plus parameters
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
4 #output : KEGG pathway : jpeg or pdf file.
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
5
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
6 suppressMessages(library("pathview"))
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
7
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
8 read_file <- function(path,header){
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
9 file <- try(read.table(path,header=header, sep="\t",stringsAsFactors = FALSE, quote=""),silent=TRUE)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
10 if (inherits(file,"try-error")){
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
11 stop("File not found !")
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
12 }else{
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
13 return(file)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
14 }
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
15 }
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
16
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
17 ##### fuction to clean and concatenate pathway name (allow more flexibility for user input)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
18 concat_string <- function(x){
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
19 x <- gsub(" - .*","",x)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
20 x <- gsub(" ","",x)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
21 x <- gsub("-","",x)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
22 x <- gsub("_","",x)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
23 x <- gsub(",","",x)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
24 x <- gsub("\\'","",x)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
25 x <- gsub("\\(.*)","",x)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
26 x <- gsub("\\/","",x)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
27 x <- tolower(x)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
28 return(x)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
29 }
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
30
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
31
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
32 get_args <- function(){
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
33
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
34 ## Collect arguments
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
35 args <- commandArgs(TRUE)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
36
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
37 ## Default setting when no arguments passed
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
38 if(length(args) < 1) {
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
39 args <- c("--help")
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
40 }
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
41
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
42 ## Help section
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
43 if("--help" %in% args) {
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
44 cat("Pathview R script
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
45 Arguments:
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
46 --help Print this test
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
47 --input path of the input file (must contains a colum of uniprot and/or geneID accession number)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
48 --id_list list of ids to use, ',' separated
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
49 --pathways_id Id(s) of pathway(s) to use, if several, semicolon separated list : hsa00010;hsa05412
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
50 --id_type Type of accession number ('uniprotID' or 'geneID')
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
51 --id_column Column containing accesion number of interest (ex : 'c1')
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
52 --header Boolean, TRUE if header FALSE if not
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
53 --ouput Output filename
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
54 --expression_values1 Column containing expression values (first condition)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
55 --expression_values2 Column containing expression values (second condition)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
56 --expression_values3 Column containing expression values (third condition)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
57 --native_kegg TRUE : native KEGG graph, FALSE : Graphviz graph
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
58 --species KEGG species (hsa, mmu, ...)
8
33c9bd75882a planemo upload commit 1953798eb28b6e45c78342ed4c6a6a978d096fca
proteore
parents: 7
diff changeset
59 --pathways_input Tab with pathways in a column, output format of find_pathways
33c9bd75882a planemo upload commit 1953798eb28b6e45c78342ed4c6a6a978d096fca
proteore
parents: 7
diff changeset
60 --pathway_col Column of pathways to use
33c9bd75882a planemo upload commit 1953798eb28b6e45c78342ed4c6a6a978d096fca
proteore
parents: 7
diff changeset
61 --header2 Boolean, TRUE if header FALSE if not
0
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
62
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
63 Example:
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
64 ./PathView.R --input 'input.csv' --pathway_id '05412' --id_type 'uniprotID' --id_column 'c1' --header TRUE \n\n")
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
65
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
66 q(save="no")
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
67 }
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
68
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
69 parseArgs <- function(x) strsplit(sub("^--", "", x), "=")
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
70 argsDF <- as.data.frame(do.call("rbind", parseArgs(args)))
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
71 args <- as.list(as.character(argsDF$V2))
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
72 names(args) <- argsDF$V1
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
73
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
74 return(args)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
75 }
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
76
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
77 str2bool <- function(x){
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
78 if (any(is.element(c("t","true"),tolower(x)))){
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
79 return (TRUE)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
80 }else if (any(is.element(c("f","false"),tolower(x)))){
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
81 return (FALSE)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
82 }else{
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
83 return(NULL)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
84 }
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
85 }
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
86
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
87 is.letter <- function(x) grepl("[[:alpha:]]", x)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
88
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
89 #### hsa00010 -> 00010
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
90 remove_kegg_prefix <- function(x){
7
b617d4bbebf8 planemo upload commit 41ba6f74dfad14f894dbe396fc765893b971b5ae-dirty
proteore
parents: 0
diff changeset
91 x = gsub(":","",x)
b617d4bbebf8 planemo upload commit 41ba6f74dfad14f894dbe396fc765893b971b5ae-dirty
proteore
parents: 0
diff changeset
92 if (substr(x,1,4) == 'path'){
b617d4bbebf8 planemo upload commit 41ba6f74dfad14f894dbe396fc765893b971b5ae-dirty
proteore
parents: 0
diff changeset
93 x=substr(x,5,nchar(x))
b617d4bbebf8 planemo upload commit 41ba6f74dfad14f894dbe396fc765893b971b5ae-dirty
proteore
parents: 0
diff changeset
94 }
0
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
95 if (is.letter(substr(x,1,3))){
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
96 x <- substr(x,4,nchar(x))
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
97 }
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
98 return(x)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
99 }
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
100
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
101 clean_bad_character <- function(string) {
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
102 string <- gsub("X","",string)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
103 string <- gsub(" ","",string)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
104 return(string)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
105 }
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
106
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
107 args <- get_args()
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
108
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
109 ###setting variables
8
33c9bd75882a planemo upload commit 1953798eb28b6e45c78342ed4c6a6a978d096fca
proteore
parents: 7
diff changeset
110 if (!is.null(args$pathways_id)) {
33c9bd75882a planemo upload commit 1953798eb28b6e45c78342ed4c6a6a978d096fca
proteore
parents: 7
diff changeset
111 ids <- sapply(rapply(strsplit(clean_bad_character(args$pathways_id),","),c), function(x) remove_kegg_prefix(x),USE.NAMES = FALSE)
33c9bd75882a planemo upload commit 1953798eb28b6e45c78342ed4c6a6a978d096fca
proteore
parents: 7
diff changeset
112 }else if (!is.null(args$pathways_input)){
33c9bd75882a planemo upload commit 1953798eb28b6e45c78342ed4c6a6a978d096fca
proteore
parents: 7
diff changeset
113 header2 <- str2bool(args$header2)
33c9bd75882a planemo upload commit 1953798eb28b6e45c78342ed4c6a6a978d096fca
proteore
parents: 7
diff changeset
114 pathway_col <- as.numeric(gsub("c", "" ,args$pathway_col))
33c9bd75882a planemo upload commit 1953798eb28b6e45c78342ed4c6a6a978d096fca
proteore
parents: 7
diff changeset
115 pathways_file = read_file(args$pathways_input,header2)
33c9bd75882a planemo upload commit 1953798eb28b6e45c78342ed4c6a6a978d096fca
proteore
parents: 7
diff changeset
116 ids <- sapply(rapply(strsplit(clean_bad_character(pathways_file[,pathway_col]),","),c), function(x) remove_kegg_prefix(x),USE.NAMES = FALSE)
33c9bd75882a planemo upload commit 1953798eb28b6e45c78342ed4c6a6a978d096fca
proteore
parents: 7
diff changeset
117 }
0
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
118 #if (!is.null(args$pathways_name)) {names <- as.vector(sapply(strsplit(args$pathways_name,","), function(x) concat_string(x),USE.NAMES = FALSE))}
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
119 if (!is.null(args$id_list)) {id_list <- as.vector(strsplit(clean_bad_character(args$id_list),","))}
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
120 id_type <- tolower(args$id_type)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
121 ncol <- as.numeric(gsub("c", "" ,args$id_column))
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
122 header <- str2bool(args$header)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
123 #output <- args$output
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
124 native_kegg <- str2bool(args$native_kegg)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
125 species=args$species
8
33c9bd75882a planemo upload commit 1953798eb28b6e45c78342ed4c6a6a978d096fca
proteore
parents: 7
diff changeset
126 #org list used in mapped2geneID
33c9bd75882a planemo upload commit 1953798eb28b6e45c78342ed4c6a6a978d096fca
proteore
parents: 7
diff changeset
127 org <- c('Hs','Mm')
33c9bd75882a planemo upload commit 1953798eb28b6e45c78342ed4c6a6a978d096fca
proteore
parents: 7
diff changeset
128 names(org) <- c('hsa','mmu')
33c9bd75882a planemo upload commit 1953798eb28b6e45c78342ed4c6a6a978d096fca
proteore
parents: 7
diff changeset
129
0
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
130
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
131
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
132 #read input file or list
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
133 if (!is.null(args$input)){
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
134 tab <- read_file(args$input,header)
7
b617d4bbebf8 planemo upload commit 41ba6f74dfad14f894dbe396fc765893b971b5ae-dirty
proteore
parents: 0
diff changeset
135 tab <- data.frame(tab[which(tab[ncol]!=""),])
0
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
136 } else {
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
137 tab <- data.frame(id_list)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
138 ncol=1
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
139 }
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
140
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
141 e1 <- as.numeric(gsub("c", "" ,args$expression_values1))
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
142 if (!is.null(args$expression_values1)) { colnames(tab)[e1] <- "e1" }
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
143 e2 <- as.numeric(gsub("c", "" ,args$expression_values2))
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
144 if (!is.null(args$expression_values2)) { colnames(tab)[e2] <- "e2" }
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
145 e3 <- as.numeric(gsub("c", "" ,args$expression_values3))
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
146 if (!is.null(args$expression_values3)) { colnames(tab)[e3] <- "e3" }
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
147
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
148
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
149 ##### map uniprotID to entrez geneID
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
150 if (id_type == "uniprotid") {
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
151
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
152 uniprotID = tab[,ncol]
8
33c9bd75882a planemo upload commit 1953798eb28b6e45c78342ed4c6a6a978d096fca
proteore
parents: 7
diff changeset
153 mapped2geneID = id2eg(ids = uniprotID, category = "uniprot", org = org[[species]], pkg.name = NULL)
0
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
154 geneID = mapped2geneID[,2]
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
155 tab = cbind(tab,geneID)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
156
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
157 }else if (id_type == "geneid"){
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
158
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
159 colnames(tab)[ncol] <- "geneID"
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
160
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
161 }
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
162
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
163 geneID = tab$geneID[which(tab$geneID !="NA")]
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
164 geneID = gsub(" ","",geneID)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
165 geneID = unlist(strsplit(geneID,"[;]"))
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
166
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
167
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
168 #### get hsa pathways list
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
169 #download.file(url = "http://rest.kegg.jp/link/pathway/hsa", destfile = "/home/dchristiany/proteore_project/ProteoRE/tools/pathview/geneID_to_hsa_pathways.csv")
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
170 #geneid_hsa_pathways <- read_file(path = "/home/dchristiany/proteore_project/ProteoRE/tools/pathview/geneID_to_hsa_pathways.csv",FALSE)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
171 #names(geneid_hsa_pathways) <- c("geneID","pathway")
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
172
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
173 ##### build matrix to map on KEGG pathway (kgml : KEGG xml)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
174 if (!is.null(args$expression_values1)&is.null(args$expression_values2)&is.null(args$expression_values3)){
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
175 mat <- as.data.frame(cbind(tab$e1)[which(!is.na(tab$geneID)),])
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
176 row.names(mat) <- tab$geneID[which(!is.na(tab$geneID))]
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
177 } else if (!is.null(args$expression_values1)&!is.null(args$expression_values2)&is.null(args$expression_values3)){
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
178 mat <- as.data.frame(cbind(tab$e1,tab$e2)[which(!is.na(tab$geneID)),])
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
179 row.names(mat) <- tab$geneID[which(!is.na(tab$geneID))]
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
180 }else if (!is.null(args$expression_values1)&!is.null(args$expression_values2)&!is.null(args$expression_values3)){
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
181 mat <- as.data.frame(cbind(tab$e1,tab$e2,tab$e3)[which(!is.na(tab$geneID)),])
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
182 row.names(mat) <- tab$geneID[which(!is.na(tab$geneID))]
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
183 } else {
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
184 mat <- geneID
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
185 }
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
186
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
187
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
188 #### simulation data test
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
189 #exp1 <- sim.mol.data(mol.type = c("gene", "gene.ko", "cpd")[1], id.type = NULL, species="hsa", discrete = FALSE, nmol = 161, nexp = 1, rand.seed=100)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
190 #exp2 <- sim.mol.data(mol.type = c("gene", "gene.ko", "cpd")[1], id.type = NULL, species="hsa", discrete = FALSE, nmol = 161, nexp = 1, rand.seed=50)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
191 #exp3 <- sim.mol.data(mol.type = c("gene", "gene.ko", "cpd")[1], id.type = NULL, species="hsa", discrete = FALSE, nmol = 161, nexp = 1, rand.seed=10)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
192 #tab <- cbind(tab,exp1,exp2,exp3)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
193
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
194 #write.table(tab, file='/home/dchristiany/proteore_project/ProteoRE/tools/pathview/Lacombe_sim_expression_data.tsv', quote=FALSE, sep='\t',row.names = FALSE)
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
195
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
196 #mat <- exp1[1:nrow(tab)]
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
197 #names(mat) <- geneID
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
198
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
199
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
200 #####mapping geneID (with or without expression values) on KEGG pathway
9
560541195c3f planemo upload commit 30488ad6185a754fd69dc983eddd171f615d26fb-dirty
proteore
parents: 8
diff changeset
201 plot.col.key= TRUE
560541195c3f planemo upload commit 30488ad6185a754fd69dc983eddd171f615d26fb-dirty
proteore
parents: 8
diff changeset
202 low_color = "green"
560541195c3f planemo upload commit 30488ad6185a754fd69dc983eddd171f615d26fb-dirty
proteore
parents: 8
diff changeset
203 mid_color = "#F3F781" #yellow
560541195c3f planemo upload commit 30488ad6185a754fd69dc983eddd171f615d26fb-dirty
proteore
parents: 8
diff changeset
204 high_color = "red"
560541195c3f planemo upload commit 30488ad6185a754fd69dc983eddd171f615d26fb-dirty
proteore
parents: 8
diff changeset
205 if (is.null(tab$e1)) {
560541195c3f planemo upload commit 30488ad6185a754fd69dc983eddd171f615d26fb-dirty
proteore
parents: 8
diff changeset
206 plot.col.key= FALSE #if there's no exrepession data, we don't show the color key
10
e8d6e50f5d38 planemo upload commit 0be58bb700f64de6792a7234a11675bae2755e8f-dirty
proteore
parents: 9
diff changeset
207 high_color = "#81BEF7" #blue
9
560541195c3f planemo upload commit 30488ad6185a754fd69dc983eddd171f615d26fb-dirty
proteore
parents: 8
diff changeset
208 }
8
33c9bd75882a planemo upload commit 1953798eb28b6e45c78342ed4c6a6a978d096fca
proteore
parents: 7
diff changeset
209
0
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
210 for (id in ids) {
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
211 pathview(gene.data = mat,
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
212 pathway.id = id,
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
213 species = species,
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
214 kegg.dir = ".",
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
215 gene.idtype = "entrez",
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
216 kegg.native = native_kegg,
9
560541195c3f planemo upload commit 30488ad6185a754fd69dc983eddd171f615d26fb-dirty
proteore
parents: 8
diff changeset
217 low = list(gene = low_color, cpd = "blue"),
560541195c3f planemo upload commit 30488ad6185a754fd69dc983eddd171f615d26fb-dirty
proteore
parents: 8
diff changeset
218 mid = list(gene = mid_color, cpd = "transparent"),
560541195c3f planemo upload commit 30488ad6185a754fd69dc983eddd171f615d26fb-dirty
proteore
parents: 8
diff changeset
219 high = list(gene = high_color, cpd = "yellow"),
10
e8d6e50f5d38 planemo upload commit 0be58bb700f64de6792a7234a11675bae2755e8f-dirty
proteore
parents: 9
diff changeset
220 na.col="#D8D8D8", #gray
9
560541195c3f planemo upload commit 30488ad6185a754fd69dc983eddd171f615d26fb-dirty
proteore
parents: 8
diff changeset
221 cpd.data=NULL,
8
33c9bd75882a planemo upload commit 1953798eb28b6e45c78342ed4c6a6a978d096fca
proteore
parents: 7
diff changeset
222 plot.col.key = plot.col.key,
0
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
223 pdf.size=c(9,9))
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
224 }
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
225
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
226 ########using keggview.native
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
227
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
228 #xml.file=system.file("extdata", "hsa00010.xml", package = "pathview")
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
229 #node.data=node.info("/home/dchristiany/hsa00010.xml")
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
230 #plot.data.gene=node.map(mol.data=test, node.data, node.types="gene")
097bb3ed2b4d planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
231 #colors =node.color(plot.data = plot.data.gene[,1:9])