annotate compute_kegg_pathways.R @ 4:7d8848c4dbb7 draft default tip

planemo upload commit 229c140e131039467e2797f656d9acf916907dd7-dirty
author proteore
date Tue, 18 Sep 2018 09:54:58 -0400
parents 0d5c21cc29f7
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
1 library(KEGGREST)
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
2
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
3 get_args <- function(){
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
4
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
5 ## Collect arguments
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
6 args <- commandArgs(TRUE)
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
7
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
8 ## Default setting when no arguments passed
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
9 if(length(args) < 1) {
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
10 args <- c("--help")
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
11 }
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
12
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
13 ## Help section
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
14 if("--help" %in% args) {
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
15 cat("Pathview R script
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
16 Arguments:
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
17 --help Print this test
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
18 --input tab file
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
19 --id_list
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
20 id list ',' separated
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
21 --id_type type of input ids (uniprot_AC or geneID)
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
22 --id_column number og column containg ids of interest
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
23 --nb_pathways number of pathways to return
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
24 --header boolean
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
25 --output output path
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
26 --ref ref file (l.hsa.gene.RData, l.hsa.up.RData, l.mmu.up.Rdata)
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
27
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
28 Example:
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
29 Rscript keggrest.R --input='P31946,P62258' --id_type='uniprot' --id_column 'c1' --header TRUE \n\n")
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
30
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
31 q(save="no")
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
32 }
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
33
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
34 parseArgs <- function(x) strsplit(sub("^--", "", x), "=")
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
35 argsDF <- as.data.frame(do.call("rbind", parseArgs(args)))
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
36 args <- as.list(as.character(argsDF$V2))
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
37 names(args) <- argsDF$V1
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
38
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
39 return(args)
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
40 }
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
41
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
42 args <- get_args()
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
43
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
44 #save(args,file="/home/dchristiany/proteore_project/ProteoRE/tools/compute_KEGG_pathways/args.Rda")
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
45 #load("/home/dchristiany/proteore_project/ProteoRE/tools/compute_KEGG_pathways/args.Rda")
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
46
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
47 ##function arguments :
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
48 ## id.ToMap = input from the user to map on the pathways = list of IDs
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
49 ## idType : must be "UNIPROT" or "ENTREZ"
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
50 ## org : for the moment can be "Hs" only. Has to evoluate to "Mm"
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
51
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
52 str2bool <- function(x){
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
53 if (any(is.element(c("t","true"),tolower(x)))){
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
54 return (TRUE)
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
55 }else if (any(is.element(c("f","false"),tolower(x)))){
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
56 return (FALSE)
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
57 }else{
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
58 return(NULL)
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
59 }
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
60 }
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
61
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
62
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
63 read_file <- function(path,header){
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
64 file <- try(read.table(path,header=header, sep="\t",stringsAsFactors = FALSE, quote=""),silent=TRUE)
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
65 if (inherits(file,"try-error")){
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
66 stop("File not found !")
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
67 }else{
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
68 return(file)
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
69 }
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
70 }
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
71
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
72 ID2KEGG.Mapping<- function(id.ToMap,ref) {
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
73
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
74 ref_ids = get(load(ref))
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
75 map<-lapply(ref_ids, is.element, unique(id.ToMap))
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
76 names(map) <- sapply(names(map), function(x) gsub("path:","",x),USE.NAMES = FALSE) #remove the prefix "path:"
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
77
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
78 in.path<-sapply(map, function(x) length(which(x==TRUE)))
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
79 tot.path<-sapply(map, length)
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
80
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
81 ratio<-(as.numeric(in.path[which(in.path!=0)])) / (as.numeric(tot.path[which(in.path!=0)]))
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
82 ratio <- as.numeric(format(round(ratio*100, 2), nsmall = 2))
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
83
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
84 ##useful but LONG
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
85 ## to do before : in step 1
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
86 path.names<-names(in.path[which(in.path!=0)])
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
87 name <- sapply(path.names, function(x) keggGet(x)[[1]]$NAME,USE.NAMES = FALSE)
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
88
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
89 res<-data.frame(I(names(in.path[which(in.path!=0)])), I(name), ratio, as.numeric(in.path[which(in.path!=0)]), as.numeric(tot.path[which(in.path!=0)]))
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
90 res <- res[order(as.numeric(res[,3]),decreasing = TRUE),]
3
0d5c21cc29f7 planemo upload commit 0be58bb700f64de6792a7234a11675bae2755e8f-dirty
proteore
parents: 0
diff changeset
91 colnames(res)<-c("pathway_ID", "Description" , "Ratio IDs mapped/total IDs (%)" ,"nb genes mapped in the pathway", "nb total genes present in the pathway")
0
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
92
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
93 return(res)
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
94
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
95 }
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
96
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
97 ###setting variables
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
98 header = str2bool(args$header)
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
99 if (!is.null(args$id_list)) {id_list <- strsplit(args$id_list,",")[[1]]}
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
100 if (!is.null(args$input)) {
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
101 csv <- read_file(args$input,header)
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
102 ncol <- as.numeric(gsub("c", "" ,args$id_column))
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
103 id_list <- as.vector(csv[,ncol])
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
104 }
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
105 id_type <- toupper(args$id_type)
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
106
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
107 #mapping on pathways
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
108 res <- ID2KEGG.Mapping(id_list,args$ref)
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
109 if (nrow(res) > as.numeric(args$nb_pathways)) { res <- res[1:args$nb_pathways,] }
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
110
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
111 write.table(res, file=args$output, quote=FALSE, sep='\t',row.names = FALSE, col.names = TRUE)
0fe3cb50fbf6 planemo upload commit 5ba22c155f8fa4a858eea9a23926c739b4722bb3-dirty
proteore
parents:
diff changeset
112