annotate GO_prof_comp.R @ 3:e337dcfb84e4 draft

planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
author proteore
date Thu, 07 Mar 2019 09:20:58 -0500
parents 67a4f68f1c1c
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
1 options(warn=-1) #TURN OFF WARNINGS !!!!!!
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
2 suppressMessages(library(clusterProfiler,quietly = TRUE))
3
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
3 suppressMessages(library(plyr, quietly = TRUE))
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
4 suppressMessages(library(ggplot2, quietly = TRUE))
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
5 suppressMessages(library(DOSE, quietly = TRUE))
0
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
6
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
7 #return the number of character from the longest description found (from the 10 first)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
8 max_str_length_10_first <- function(vector){
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
9 vector <- as.vector(vector)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
10 nb_description = length(vector)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
11 if (nb_description >= 10){nb_description=10}
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
12 return(max(nchar(vector[1:nb_description])))
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
13 }
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
14
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
15 str2bool <- function(x){
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
16 if (any(is.element(c("t","true"),tolower(x)))){
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
17 return (TRUE)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
18 }else if (any(is.element(c("f","false"),tolower(x)))){
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
19 return (FALSE)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
20 }else{
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
21 return(NULL)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
22 }
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
23 }
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
24
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
25 get_args <- function(){
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
26
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
27 ## Collect arguments
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
28 args <- commandArgs(TRUE)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
29
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
30 ## Default setting when no arguments passed
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
31 if(length(args) < 1) {
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
32 args <- c("--help")
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
33 }
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
34
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
35 ## Help section
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
36 if("--help" %in% args) {
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
37 cat("Selection and Annotation HPA
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
38 Arguments:
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
39 --inputtype1: type of input (list of id or filename)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
40 --inputtype2: type of input (list of id or filename)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
41 --input1: input1
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
42 --input2: input2
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
43 --column1: the column number which you would like to apply...
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
44 --column2: the column number which you would like to apply...
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
45 --header1: true/false if your file contains a header
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
46 --header2: true/false if your file contains a header
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
47 --ont: ontology to use
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
48 --lev: ontology level
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
49 --org: organism db package
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
50 --list_name1: name of the first list
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
51 --list_name2: name of the second list \n")
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
52
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
53 q(save="no")
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
54 }
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
55
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
56 parseArgs <- function(x) strsplit(sub("^--", "", x), "=")
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
57 argsDF <- as.data.frame(do.call("rbind", parseArgs(args)))
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
58 args <- as.list(as.character(argsDF$V2))
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
59 names(args) <- argsDF$V1
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
60
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
61 return(args)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
62 }
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
63
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
64 get_ids=function(inputtype, input, ncol, header) {
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
65
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
66 if (inputtype == "text") {
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
67 ids = strsplit(input, "[ \t\n]+")[[1]]
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
68 } else if (inputtype == "file") {
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
69 header=str2bool(header)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
70 ncol=get_cols(ncol)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
71 csv = read.csv(input,header=header, sep="\t", as.is=T)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
72 ids=csv[,ncol]
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
73 }
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
74
2
67a4f68f1c1c planemo upload commit 12b0b3e18cd7af9a8588906fa2ec199b9d2a770f-dirty
proteore
parents: 0
diff changeset
75 ids = unlist(strsplit(as.character(ids),";"))
0
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
76 ids = ids[which(!is.na(ids))]
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
77
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
78 return(ids)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
79 }
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
80
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
81 str2bool <- function(x){
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
82 if (any(is.element(c("t","true"),tolower(x)))){
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
83 return (TRUE)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
84 }else if (any(is.element(c("f","false"),tolower(x)))){
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
85 return (FALSE)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
86 }else{
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
87 return(NULL)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
88 }
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
89 }
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
90
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
91 check_ids <- function(vector,type) {
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
92 uniprot_pattern = "^([OPQ][0-9][A-Z0-9]{3}[0-9]|[A-NR-Z][0-9]([A-Z][A-Z0-9]{2}[0-9]){1,2})$"
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
93 entrez_id = "^([0-9]+|[A-Z]{1,2}_[0-9]+|[A-Z]{1,2}_[A-Z]{1,4}[0-9]+)$"
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
94 if (type == "entrez")
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
95 return(grepl(entrez_id,vector))
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
96 else if (type == "uniprot") {
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
97 return(grepl(uniprot_pattern,vector))
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
98 }
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
99 }
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
100
3
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
101 #res.cmp@compareClusterResult$Description <- sapply(as.vector(res.cmp@compareClusterResult$Description), function(x) {ifelse(nchar(x)>50, substr(x,1,50),x)},USE.NAMES = FALSE)
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
102 fortify.compareClusterResult <- function(res.cmp, showCategory=30, by="geneRatio", split=NULL, includeAll=TRUE) {
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
103 clProf.df <- as.data.frame(res.cmp)
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
104 .split <- split
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
105 ## get top 5 (default) categories of each gene cluster.
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
106 if (is.null(showCategory)) {
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
107 result <- clProf.df
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
108 } else {
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
109 Cluster <- NULL # to satisfy codetools
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
110 topN <- function(res, showCategory) {
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
111 ddply(.data = res, .variables = .(Cluster), .fun = function(df, N) {
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
112 if (length(df$Count) > N) {
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
113 if (any(colnames(df) == "pvalue")) {
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
114 idx <- order(df$pvalue, decreasing=FALSE)[1:N]
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
115 } else {
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
116 ## for groupGO
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
117 idx <- order(df$Count, decreasing=T)[1:N]
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
118 }
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
119 return(df[idx,])
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
120 } else {
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
121 return(df)
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
122 }
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
123 },
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
124 N=showCategory
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
125 )
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
126 }
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
127 if (!is.null(.split) && .split %in% colnames(clProf.df)) {
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
128 lres <- split(clProf.df, as.character(clProf.df[, .split]))
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
129 lres <- lapply(lres, topN, showCategory = showCategory)
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
130 result <- do.call('rbind', lres)
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
131 } else {
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
132 result <- topN(clProf.df, showCategory)
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
133 }
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
134 }
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
135 ID <- NULL
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
136 if (includeAll == TRUE) {
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
137 result = subset(clProf.df, ID %in% result$ID)
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
138 }
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
139 ## remove zero count
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
140 result$Description <- as.character(result$Description) ## un-factor
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
141 GOlevel <- result[,c("ID", "Description")] ## GO ID and Term
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
142 GOlevel <- unique(GOlevel)
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
143 result <- result[result$Count != 0, ]
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
144 result$Description <- factor(result$Description,levels=rev(GOlevel[,2]))
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
145 if (by=="rowPercentage") {
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
146 Description <- Count <- NULL # to satisfy codetools
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
147 result <- ddply(result,.(Description),transform,Percentage = Count/sum(Count),Total = sum(Count))
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
148 ## label GO Description with gene counts.
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
149 x <- mdply(result[, c("Description", "Total")], paste, sep=" (")
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
150 y <- sapply(x[,3], paste, ")", sep="")
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
151 result$Description <- y
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
152
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
153 ## restore the original order of GO Description
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
154 xx <- result[,c(2,3)]
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
155 xx <- unique(xx)
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
156 rownames(xx) <- xx[,1]
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
157 Termlevel <- xx[as.character(GOlevel[,1]),2]
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
158
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
159 ##drop the *Total* column
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
160 result <- result[, colnames(result) != "Total"]
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
161 result$Description <- factor(result$Description, levels=rev(Termlevel))
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
162
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
163 } else if (by == "count") {
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
164 ## nothing
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
165 } else if (by == "geneRatio") { ##default
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
166 gsize <- as.numeric(sub("/\\d+$", "", as.character(result$GeneRatio)))
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
167 gcsize <- as.numeric(sub("^\\d+/", "", as.character(result$GeneRatio)))
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
168 result$GeneRatio = gsize/gcsize
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
169 cluster <- paste(as.character(result$Cluster),"\n", "(", gcsize, ")", sep="")
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
170 lv <- unique(cluster)[order(as.numeric(unique(result$Cluster)))]
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
171 result$Cluster <- factor(cluster, levels = lv)
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
172 } else {
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
173 ## nothing
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
174 }
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
175 return(result)
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
176 }
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
177
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
178 ##function plotting.clusteProfile from clusterProfiler pkg
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
179 plotting.clusterProfile <- function(clProf.reshape.df,x = ~Cluster,type = "dot", colorBy = "p.adjust",by = "geneRatio",title="",font.size=12) {
0
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
180
3
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
181 Description <- Percentage <- Count <- Cluster <- GeneRatio <- p.adjust <- pvalue <- NULL # to
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
182 if (type == "dot") {
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
183 if (by == "rowPercentage") {
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
184 p <- ggplot(clProf.reshape.df,
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
185 aes_(x = x, y = ~Description, size = ~Percentage))
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
186 } else if (by == "count") {
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
187 p <- ggplot(clProf.reshape.df,
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
188 aes_(x = x, y = ~Description, size = ~Count))
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
189 } else if (by == "geneRatio") { ##DEFAULT
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
190 p <- ggplot(clProf.reshape.df,
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
191 aes_(x = x, y = ~Description, size = ~GeneRatio))
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
192 } else {
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
193 ## nothing here
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
194 }
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
195 if (any(colnames(clProf.reshape.df) == colorBy)) {
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
196 p <- p +
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
197 geom_point() +
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
198 aes_string(color=colorBy) +
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
199 scale_color_continuous(low="red", high="blue", guide=guide_colorbar(reverse=TRUE))
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
200 ## scale_color_gradientn(guide=guide_colorbar(reverse=TRUE), colors = enrichplot:::sig_palette)
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
201 } else {
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
202 p <- p + geom_point(colour="steelblue")
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
203 }
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
204 }
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
205
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
206 p <- p + xlab("") + ylab("") + ggtitle(title) +
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
207 theme_dose(font.size)
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
208
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
209 ## theme(axis.text.x = element_text(colour="black", size=font.size, vjust = 1)) +
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
210 ## theme(axis.text.y = element_text(colour="black",
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
211 ## size=font.size, hjust = 1)) +
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
212 ## ggtitle(title)+theme_bw()
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
213 ## p <- p + theme(axis.text.x = element_text(angle=angle.axis.x,
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
214 ## hjust=hjust.axis.x,
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
215 ## vjust=vjust.axis.x))
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
216
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
217 return(p)
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
218 }
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
219
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
220 make_dotplot<-function(res.cmp,ontology) {
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
221
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
222 dfok<-fortify.compareClusterResult(res.cmp)
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
223 dfok$Description <- sapply(as.vector(dfok$Description), function(x) {ifelse(nchar(x)>50, substr(x,1,50),x)},USE.NAMES = FALSE)
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
224 p<-plotting.clusterProfile(dfok, title="")
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
225
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
226 #plot(p, type="dot") #
0
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
227 output_path= paste("GO_profiles_comp_",ontology,".png",sep="")
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
228 png(output_path,height = 720, width = 600)
3
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
229 pl <- plot(p, type="dot")
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
230 print(pl)
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
231 dev.off()
0
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
232 }
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
233
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
234 get_cols <-function(input_cols) {
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
235 input_cols <- gsub("c","",gsub("C","",gsub(" ","",input_cols)))
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
236 if (grepl(":",input_cols)) {
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
237 first_col=unlist(strsplit(input_cols,":"))[1]
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
238 last_col=unlist(strsplit(input_cols,":"))[2]
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
239 cols=first_col:last_col
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
240 } else {
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
241 cols = as.integer(unlist(strsplit(input_cols,",")))
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
242 }
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
243 return(cols)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
244 }
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
245
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
246 #to check
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
247 cmp.GO <- function(l,fun="groupGO",orgdb, ontology, level=3, readable=TRUE) {
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
248 cmpGO<-compareCluster(geneClusters = l,
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
249 fun=fun,
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
250 OrgDb = orgdb,
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
251 ont=ontology,
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
252 level=level,
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
253 readable=TRUE)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
254
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
255 return(cmpGO)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
256 }
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
257
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
258 check_ids <- function(vector,type) {
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
259 uniprot_pattern = "^([OPQ][0-9][A-Z0-9]{3}[0-9]|[A-NR-Z][0-9]([A-Z][A-Z0-9]{2}[0-9]){1,2})$"
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
260 entrez_id = "^([0-9]+|[A-Z]{1,2}_[0-9]+|[A-Z]{1,2}_[A-Z]{1,4}[0-9]+)$"
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
261 if (type == "entrez")
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
262 return(grepl(entrez_id,vector))
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
263 else if (type == "uniprot") {
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
264 return(grepl(uniprot_pattern,vector))
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
265 }
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
266 }
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
267
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
268 main = function() {
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
269
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
270 #to get the args of the command line
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
271 args=get_args()
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
272
3
e337dcfb84e4 planemo upload commit 51fc514a85c1055cab5bb6e76c90f3da7e648101-dirty
proteore
parents: 2
diff changeset
273
0
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
274 ids1<-get_ids(args$inputtype1, args$input1, args$column1, args$header1)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
275 ids2<-get_ids(args$inputtype2, args$input2, args$column2, args$header2)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
276 ont = strsplit(args$ont, ",")[[1]]
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
277 lev=as.integer(args$lev)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
278 org=args$org
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
279
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
280 #load annot package
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
281 suppressMessages(library(args$org, character.only = TRUE, quietly = TRUE))
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
282
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
283 # Extract OrgDb
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
284 if (args$org=="org.Hs.eg.db") {
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
285 orgdb<-org.Hs.eg.db
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
286 } else if (args$org=="org.Mm.eg.db") {
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
287 orgdb<-org.Mm.eg.db
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
288 } else if (args$org=="org.Rn.eg.db") {
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
289 orgdb<-org.Rn.eg.db
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
290 }
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
291
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
292 for(ontology in ont) {
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
293 liste = list("l1"=ids1,"l2"=ids2)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
294 names(liste) = c(args$list_name1,args$list_name2)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
295 res.cmp<-cmp.GO(l=liste,fun="groupGO",orgdb, ontology, level=lev, readable=TRUE)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
296 make_dotplot(res.cmp,ontology)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
297 output_path = paste("GO_profiles_comp_",ontology,".tsv",sep="")
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
298 write.table(res.cmp@compareClusterResult, output_path, sep="\t", row.names=F, quote=F)
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
299 }
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
300
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
301 } #end main
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
302
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
303 main()
a0ea9e00bd66 planemo upload commit e5e4a3efd1cb1498724c0e93f8ec9606d0ed3045-dirty
proteore
parents:
diff changeset
304