annotate keggrest.R @ 1:b9bf788be4dd draft

planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
author proteore
date Wed, 22 Aug 2018 11:16:00 -0400
parents 4da8523282cf
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
1 library(KEGGREST)
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
2
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
3 get_args <- function(){
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
4
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
5 ## Collect arguments
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
6 args <- commandArgs(TRUE)
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
7
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
8 ## Default setting when no arguments passed
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
9 if(length(args) < 1) {
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
10 args <- c("--help")
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
11 }
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
12
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
13 ## Help section
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
14 if("--help" %in% args) {
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
15 cat("Pathview R script
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
16 Arguments:
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
17 --help Print this test
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
18 --input tab file
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
19 --id_list
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
20 id list ',' separated
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
21 --id_type type of input ids (uniprot_AC or geneID)
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
22 --id_column number og column containg ids of interest
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
23 --nb_pathways number of pathways to return
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
24 --header boolean
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
25 --output output path
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
26 --ref ref file (l.hsa.gene.RData, l.hsa.up.RData, l.mmu.up.Rdata)
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
27
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
28 Example:
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
29 Rscript keggrest.R --input='P31946,P62258' --id_type='uniprot' --id_column 'c1' --header TRUE \n\n")
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
30
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
31 q(save="no")
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
32 }
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
33
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
34
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
35 #save(args,file="/home/dchristiany/proteore_project/ProteoRE/tools/find_pathways/args.Rda")
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
36 #load("/home/dchristiany/proteore_project/ProteoRE/tools/find_pathways/args.Rda")
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
37 parseArgs <- function(x) strsplit(sub("^--", "", x), "=")
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
38 argsDF <- as.data.frame(do.call("rbind", parseArgs(args)))
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
39 args <- as.list(as.character(argsDF$V2))
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
40 names(args) <- argsDF$V1
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
41
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
42 return(args)
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
43 }
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
44
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
45 args <- get_args()
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
46
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
47 ##function arguments :
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
48 ## id.ToMap = input from the user to map on the pathways = list of IDs
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
49 ## idType : must be "UNIPROT" or "ENTREZ"
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
50 ## org : for the moment can be "Hs" only. Has to evoluate to "Mm"
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
51
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
52 str2bool <- function(x){
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
53 if (any(is.element(c("t","true"),tolower(x)))){
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
54 return (TRUE)
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
55 }else if (any(is.element(c("f","false"),tolower(x)))){
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
56 return (FALSE)
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
57 }else{
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
58 return(NULL)
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
59 }
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
60 }
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
61
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
62
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
63 read_file <- function(path,header){
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
64 file <- try(read.table(path,header=header, sep="\t",stringsAsFactors = FALSE, quote=""),silent=TRUE)
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
65 if (inherits(file,"try-error")){
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
66 stop("File not found !")
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
67 }else{
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
68 return(file)
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
69 }
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
70 }
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
71
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
72 ID2KEGG.Mapping<- function(id.ToMap,ref) {
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
73
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
74 ref_ids = get(load(ref))
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
75 map<-lapply(ref_ids, is.element, unique(id.ToMap))
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
76
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
77 in.path<-sapply(map, function(x) length(which(x==TRUE)))
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
78 tot.path<-sapply(map, length)
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
79
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
80 ratio<-(as.numeric(in.path[which(in.path!=0)])) / (as.numeric(tot.path[which(in.path!=0)]))
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
81 ratio <- as.numeric(format(round(ratio*100, 2), nsmall = 2))
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
82
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
83 ##useful but LONG
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
84 ## to do before : in step 1
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
85 path.names<-names(in.path[which(in.path!=0)])
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
86 name <- sapply(path.names, function(x) keggGet(x)[[1]]$NAME,USE.NAMES = FALSE)
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
87
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
88 res<-data.frame(I(names(in.path[which(in.path!=0)])), I(name), ratio, as.numeric(in.path[which(in.path!=0)]), as.numeric(tot.path[which(in.path!=0)]))
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
89 res <- res[order(as.numeric(res[,3]),decreasing = TRUE),]
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
90 colnames(res)<-c("pathID", "Desc" , "ratio mapped/total (percent)" ,"genes mapped in the path", "tot genes in the path")
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
91
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
92 return(res)
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
93
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
94 }
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
95
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
96 ###setting variables
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
97 header = str2bool(args$header)
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
98 if (!is.null(args$id_list)) {id_list <- strsplit(args$id_list,",")[[1]]}
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
99 if (!is.null(args$input)) {
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
100 csv <- read_file(args$input,header)
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
101 ncol <- as.numeric(gsub("c", "" ,args$id_column))
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
102 id_list <- as.vector(csv[,ncol])
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
103 }
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
104 id_type <- toupper(args$id_type)
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
105
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
106 #mapping on pathways
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
107 res <- ID2KEGG.Mapping(id_list,args$ref)
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
108 if (nrow(res) > as.numeric(args$nb_pathways)) { res <- res[1:args$nb_pathways,] }
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
109
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
110 write.table(res, file=args$output, quote=FALSE, sep='\t',row.names = FALSE, col.names = TRUE)
4da8523282cf planemo upload commit 2e441b4969ae7cf9aeb227a1d47c43ef7268a5e6-dirty
proteore
parents:
diff changeset
111