Mercurial > repos > proteore > proteore_kegg_pathways_coverage
annotate compute_kegg_pathways.R @ 0:8883a7173cba draft
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
| author | proteore |
|---|---|
| date | Mon, 12 Nov 2018 10:59:49 -0500 |
| parents | |
| children | 7004924a3686 |
| rev | line source |
|---|---|
|
0
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
1 options(warn=-1) #TURN OFF WARNINGS !!!!!! |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
2 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
3 suppressMessages(library(KEGGREST)) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
4 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
5 get_args <- function(){ |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
6 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
7 ## Collect arguments |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
8 args <- commandArgs(TRUE) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
9 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
10 ## Default setting when no arguments passed |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
11 if(length(args) < 1) { |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
12 args <- c("--help") |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
13 } |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
14 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
15 ## Help section |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
16 if("--help" %in% args) { |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
17 cat("Pathview R script |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
18 Arguments: |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
19 --help Print this test |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
20 --input tab file |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
21 --id_list id list ',' separated |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
22 --id_type type of input ids (kegg-id, uniprot_AC,geneID) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
23 --id_column number og column containg ids of interest |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
24 --nb_pathways number of pathways to return |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
25 --header boolean |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
26 --output output path |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
27 --species species used to get specific pathways (hsa,mmu,rno) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
28 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
29 Example: |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
30 Rscript keggrest.R --input='P31946,P62258' --id_type='uniprot' --id_column 'c1' --header TRUE \n\n") |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
31 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
32 q(save="no") |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
33 } |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
34 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
35 parseArgs <- function(x) strsplit(sub("^--", "", x), "=") |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
36 argsDF <- as.data.frame(do.call("rbind", parseArgs(args))) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
37 args <- as.list(as.character(argsDF$V2)) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
38 names(args) <- argsDF$V1 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
39 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
40 return(args) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
41 } |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
42 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
43 str2bool <- function(x){ |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
44 if (any(is.element(c("t","true"),tolower(x)))){ |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
45 return (TRUE) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
46 }else if (any(is.element(c("f","false"),tolower(x)))){ |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
47 return (FALSE) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
48 }else{ |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
49 return(NULL) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
50 } |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
51 } |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
52 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
53 read_file <- function(path,header){ |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
54 file <- try(read.csv(path,header=header, sep="\t",stringsAsFactors = FALSE, quote="\"", check.names = F),silent=TRUE) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
55 if (inherits(file,"try-error")){ |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
56 stop("File not found !") |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
57 }else{ |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
58 return(file) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
59 } |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
60 } |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
61 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
62 get_pathways_list <- function(species){ |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
63 ##all available pathways for the species |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
64 pathways <-keggLink("pathway", species) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
65 tot_path<-unique(pathways) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
66 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
67 ##formating the dat into a list object |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
68 ##key= pathway ID, value = genes of the pathway in the kegg format |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
69 pathways_list <- sapply(tot_path, function(pathway) names(which(pathways==pathway))) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
70 return (pathways_list) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
71 } |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
72 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
73 get_list_from_cp <-function(list){ |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
74 list = strsplit(list, "[ \t\n]+")[[1]] |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
75 list = list[list != ""] #remove empty entry |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
76 list = gsub("-.+", "", list) #Remove isoform accession number (e.g. "-2") |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
77 return(list) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
78 } |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
79 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
80 geneID_to_kegg <- function(vector,species){ |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
81 vector <- sapply(vector, function(x) paste(species,x,sep=":"),USE.NAMES = F) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
82 return (vector) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
83 } |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
84 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
85 kegg_mapping<- function(kegg_id_list,id_type,ref_ids) { |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
86 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
87 #convert to KEGG ID |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
88 #if (id_type!="kegg-id"){ |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
89 # id_list <- unique(sapply(id_list, function(x) paste(id_type,":",x,sep=""),USE.NAMES = F)) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
90 # if (length(id_list)>250){ |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
91 # id_list <- split(id_list, ceiling(seq_along(id_list)/250)) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
92 # id_list <- sapply(id_list, function(x) keggConv("genes",x)) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
93 # kegg_id_list <- unique(unlist(id_list)) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
94 # } else { |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
95 # kegg_id_list <- unique(keggConv("genes", id_list)) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
96 # } |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
97 #} else { |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
98 # kegg_id_list <- unique(id_list) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
99 #} |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
100 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
101 #mapping |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
102 map<-lapply(ref_ids, is.element, unique(kegg_id_list)) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
103 names(map) <- sapply(names(map), function(x) gsub("path:","",x),USE.NAMES = FALSE) #remove the prefix "path:" |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
104 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
105 in.path<-sapply(map, function(x) length(which(x==TRUE))) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
106 tot.path<-sapply(map, length) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
107 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
108 ratio <- (as.numeric(in.path[which(in.path!=0)])) / (as.numeric(tot.path[which(in.path!=0)])) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
109 ratio <- as.numeric(format(round(ratio*100, 2), nsmall = 2)) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
110 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
111 ##useful but LONG |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
112 ## to do before : in step 1 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
113 path.names<-names(in.path[which(in.path!=0)]) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
114 name <- sapply(path.names, function(x) keggGet(x)[[1]]$NAME,USE.NAMES = FALSE) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
115 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
116 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)])) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
117 res <- res[order(as.numeric(res[,3]),decreasing = TRUE),] |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
118 colnames(res)<-c("pathway_ID", "Description" , "Ratio IDs mapped/total IDs (%)" ,"nb KEGG genes IDs mapped in the pathway", "nb total of KEGG genes IDs present in the pathway") |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
119 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
120 return(res) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
121 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
122 } |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
123 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
124 #get args from command line |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
125 args <- get_args() |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
126 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
127 #save(args,file="/home/dchristiany/proteore_project/ProteoRE/tools/kegg_pathways_identification/args.Rda") |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
128 #load("/home/dchristiany/proteore_project/ProteoRE/tools/kegg_pathways_identification/args.Rda") |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
129 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
130 ###setting variables |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
131 header = str2bool(args$header) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
132 if (!is.null(args$id_list)) {id_list <- get_list_from_cp(args$id_list)} |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
133 if (!is.null(args$input)) { |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
134 csv <- read_file(args$input,header) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
135 ncol <- as.numeric(gsub("c", "" ,args$id_column)) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
136 id_list <- as.vector(csv[,ncol]) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
137 id_list <- id_list[which(!is.na(id_list))] |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
138 } |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
139 if (args$id_type == "ncbi-geneid") { |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
140 id_list <- geneID_to_kegg(id_list,args$species) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
141 } |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
142 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
143 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
144 #get pathways of species with associated KEGG ID genes |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
145 pathways_list <- get_pathways_list(args$species) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
146 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
147 #mapping on pathways |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
148 res <- kegg_mapping(id_list,args$id_type,pathways_list) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
149 if (nrow(res) > as.numeric(args$nb_pathways)) { res <- res[1:args$nb_pathways,] } |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
150 |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
151 write.table(res, file=args$output, quote=FALSE, sep='\t',row.names = FALSE, col.names = TRUE) |
|
8883a7173cba
planemo upload commit 63302cb49d4f0f4dbc9ae141d20704822588f54e-dirty
proteore
parents:
diff
changeset
|
152 |
