annotate compute_kegg_pathways.R @ 0:42d0805353b6 draft

planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
author proteore
date Wed, 19 Sep 2018 05:38:52 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
1 library(KEGGREST)
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
2
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
3 get_args <- function(){
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
4
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
5 ## Collect arguments
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
6 args <- commandArgs(TRUE)
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
7
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
8 ## Default setting when no arguments passed
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
9 if(length(args) < 1) {
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
10 args <- c("--help")
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
11 }
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
12
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
13 ## Help section
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
14 if("--help" %in% args) {
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
15 cat("Pathview R script
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
16 Arguments:
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
17 --help Print this test
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
18 --input tab file
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
19 --id_list
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
20 id list ',' separated
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
21 --id_type type of input ids (uniprot_AC or geneID)
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
22 --id_column number og column containg ids of interest
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
23 --nb_pathways number of pathways to return
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
24 --header boolean
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
25 --output output path
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
26 --ref ref file (l.hsa.gene.RData, l.hsa.up.RData, l.mmu.up.Rdata)
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
27
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
28 Example:
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
29 Rscript keggrest.R --input='P31946,P62258' --id_type='uniprot' --id_column 'c1' --header TRUE \n\n")
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
30
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
31 q(save="no")
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
32 }
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
33
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
34 parseArgs <- function(x) strsplit(sub("^--", "", x), "=")
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
35 argsDF <- as.data.frame(do.call("rbind", parseArgs(args)))
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
36 args <- as.list(as.character(argsDF$V2))
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
37 names(args) <- argsDF$V1
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
38
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
39 return(args)
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
40 }
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
41
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
42 args <- get_args()
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
43
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
44 #save(args,file="/home/dchristiany/proteore_project/ProteoRE/tools/compute_KEGG_pathways/args.Rda")
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
45 #load("/home/dchristiany/proteore_project/ProteoRE/tools/compute_KEGG_pathways/args.Rda")
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
46
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
47 ##function arguments :
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
48 ## id.ToMap = input from the user to map on the pathways = list of IDs
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
49 ## idType : must be "UNIPROT" or "ENTREZ"
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
50 ## org : for the moment can be "Hs" only. Has to evoluate to "Mm"
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
51
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
52 str2bool <- function(x){
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
53 if (any(is.element(c("t","true"),tolower(x)))){
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
54 return (TRUE)
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
55 }else if (any(is.element(c("f","false"),tolower(x)))){
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
56 return (FALSE)
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
57 }else{
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
58 return(NULL)
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
59 }
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
60 }
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
61
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
62
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
63 read_file <- function(path,header){
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
64 file <- try(read.table(path,header=header, sep="\t",stringsAsFactors = FALSE, quote=""),silent=TRUE)
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
65 if (inherits(file,"try-error")){
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
66 stop("File not found !")
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
67 }else{
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
68 return(file)
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
69 }
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
70 }
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
71
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
72 ID2KEGG.Mapping<- function(id.ToMap,ref) {
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
73
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
74 ref_ids = get(load(ref))
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
75 map<-lapply(ref_ids, is.element, unique(id.ToMap))
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
76 names(map) <- sapply(names(map), function(x) gsub("path:","",x),USE.NAMES = FALSE) #remove the prefix "path:"
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
77
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
78 in.path<-sapply(map, function(x) length(which(x==TRUE)))
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
79 tot.path<-sapply(map, length)
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
80
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
81 ratio<-(as.numeric(in.path[which(in.path!=0)])) / (as.numeric(tot.path[which(in.path!=0)]))
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
82 ratio <- as.numeric(format(round(ratio*100, 2), nsmall = 2))
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
83
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
84 ##useful but LONG
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
85 ## to do before : in step 1
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
86 path.names<-names(in.path[which(in.path!=0)])
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
87 name <- sapply(path.names, function(x) keggGet(x)[[1]]$NAME,USE.NAMES = FALSE)
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
88
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
89 res<-data.frame(I(names(in.path[which(in.path!=0)])), I(name), ratio, as.numeric(in.path[which(in.path!=0)]), as.numeric(tot.path[which(in.path!=0)]))
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
90 res <- res[order(as.numeric(res[,3]),decreasing = TRUE),]
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
91 colnames(res)<-c("pathway_ID", "Description" , "Ratio IDs mapped/total IDs (%)" ,"nb genes mapped in the pathway", "nb total genes present in the pathway")
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
92
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
93 return(res)
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
94
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
95 }
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
96
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
97 ###setting variables
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
98 header = str2bool(args$header)
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
99 if (!is.null(args$id_list)) {id_list <- strsplit(args$id_list,",")[[1]]}
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
100 if (!is.null(args$input)) {
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
101 csv <- read_file(args$input,header)
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
102 ncol <- as.numeric(gsub("c", "" ,args$id_column))
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
103 id_list <- as.vector(csv[,ncol])
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
104 }
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
105 id_type <- toupper(args$id_type)
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
106
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
107 #mapping on pathways
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
108 res <- ID2KEGG.Mapping(id_list,args$ref)
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
109 if (nrow(res) > as.numeric(args$nb_pathways)) { res <- res[1:args$nb_pathways,] }
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
110
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
111 write.table(res, file=args$output, quote=FALSE, sep='\t',row.names = FALSE, col.names = TRUE)
42d0805353b6 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents:
diff changeset
112