Mercurial > repos > proteore > proteore_goprofiles

annotate goprofiles.R @ 15:601027649251 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
author proteore
date Tue, 06 Apr 2021 16:59:30 +0000
parents 3ddc1f78773d
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
1 options(warn = -1) #TURN OFF WARNINGS !!!!!!
6
6afe8166a9a4 planemo upload commit 4e898239997b7ef266b1d0ce0a4c7cdec31b4ecd-dirty
proteore
parents: 5
diff changeset
2
0
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
3 # Load necessary libraries
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
4 suppressMessages(library(goProfiles, quietly = TRUE))
0
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
5
4
715002a394ec planemo upload commit b36435833bf54f90f62cc240f2cda1c889161b23-dirty
proteore
parents: 2
diff changeset
6 # Read file and return file content as data.frame
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
7 read_file <- function(path, header) {
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
8 file <- try(read.csv(path, header = header,
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
9 sep = "\t", stringsAsFactors = FALSE,
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
10 quote = "\"", check.names = F), silent = TRUE)
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
11 if (inherits(file, "try-error")) {
8
386145573c19 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents: 7
diff changeset
12 stop("File not found !")
386145573c19 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents: 7
diff changeset
13 }else{
386145573c19 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents: 7
diff changeset
14 return(file)
0
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
15 }
8
386145573c19 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents: 7
diff changeset
16 }
386145573c19 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents: 7
diff changeset
17
386145573c19 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents: 7
diff changeset
18 #convert a string to boolean
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
19 str2bool <- function(x) {
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
20 if (any(is.element(c("t", "true"), tolower(x)))) {
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
21 return(TRUE)
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
22 }else if (any(is.element(c("f", "false"), tolower(x)))) {
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
23 return(FALSE)
8
386145573c19 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents: 7
diff changeset
24 }else{
386145573c19 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents: 7
diff changeset
25 return(NULL)
0
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
26 }
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
27 }
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
28
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
29 check_ids <- function(vector, type) {
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
30 # nolint start
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
31 uniprot_pattern <-
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
32 "^([OPQ][0-9][A-Z0-9]{3}[0-9]|[A-NR-Z][0-9]([A-Z][A-Z0-9]{2}[0-9]){1,2})$"
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
33 entrez_id <- "^([0-9]+|[A-Z]{1,2}_[0-9]+|[A-Z]{1,2}_[A-Z]{1,4}[0-9]+)$"
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
34 # nolint end
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
35 if (type == "Entrez") {
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
36 return(grepl(entrez_id, vector))
5
781072a65600 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents: 4
diff changeset
37 } else if (type == "UniProt") {
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
38 return(grepl(uniprot_pattern, vector))
5
781072a65600 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents: 4
diff changeset
39 }
781072a65600 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents: 4
diff changeset
40 }
781072a65600 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents: 4
diff changeset
41
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
42 getprofile <- function(ids, id_type, level, duplicate, species) {
0
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
43 ####################################################################
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
44 # Arguments
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
45 # - ids: list of input IDs
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
46 # - id_type: type of input IDs (UniProt/ENTREZID)
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
47 # - level
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
48 # - duplicate: if the duplicated IDs should be removed or not (TRUE/FALSE)
5
781072a65600 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents: 4
diff changeset
49 # - species
0
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
50 ####################################################################
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
51
5
781072a65600 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents: 4
diff changeset
52 library(species, character.only = TRUE, quietly = TRUE)
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
53
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
54 if (species == "org.Hs.eg.db") {
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
55 package <- org.Hs.eg.db #nolint
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
56 } else if (species == "org.Mm.eg.db") {
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
57 package <- org.Mm.eg.db #nolint
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
58 } else if (species == "org.Rn.eg.db") {
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
59 package <- org.Rn.eg.db #nolint
5
781072a65600 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents: 4
diff changeset
60 }
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
61
0
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
62 # Check if level is number
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
63 if (! as.numeric(level) %% 1 == 0) {
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
64 stop("Please enter an integer for level")
5
781072a65600 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents: 4
diff changeset
65 } else {
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
66 level <- as.numeric(level)
0
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
67 }
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
68
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
69
0
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
70 # Extract Gene Entrez ID
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
71 if (id_type == "Entrez") {
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
72 id <- select(package, ids, "ENTREZID", multiVals = "first") #nolint
5
781072a65600 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents: 4
diff changeset
73 } else {
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
74 id <- select(package, ids, "ENTREZID", "UNIPROT", multiVals = "first") #nolint
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
75 }
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
76 if (duplicate) {
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
77 id <- unique(id)
0
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
78 }
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
79 genes_ids <- id$ENTREZID[which(!is.na(id$ENTREZID))]
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
80 nas <- id$UNIPROT[which(is.na(id$ENTREZID))] # IDs that have NA ENTREZID #nolint
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
81
0
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
82 # Create basic profiles
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
83 profile.CC <-
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
84 basicProfile(genes_ids, onto = "CC", level = level,
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
85 orgPackage = species, empty.cats = F, ord = T, na.rm = T)
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
86 profile.BP <-
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
87 basicProfile(genes_ids, onto = "BP", level = level,
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
88 orgPackage = species, empty.cats = F, ord = T, na.rm = T)
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
89 profile.MF <-
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
90 basicProfile(genes_ids, onto = "MF", level = level,
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
91 orgPackage = species, empty.cats = F, ord = T, na.rm = T)
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
92 profile.ALL <-
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
93 basicProfile(genes_ids, onto = "ANY", level = level,
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
94 orgPackage = species, empty.cats = F, ord = T, na.rm = T)
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
95
0
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
96 return(c(profile.CC, profile.MF, profile.BP, profile.ALL))
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
97 }
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
98
9
948fecb6a40b planemo upload commit 973d782455fd6a7df7ba9bce0a7878de53bc1e68-dirty
proteore
parents: 8
diff changeset
99 #return height and width of plot in inches from profile
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
100 plot_size_from_nb_onto <- function(profile) {
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
101 width <- 10
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
102 range <- seq(25, 2000, by = 25)
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
103 names(range) <- seq(5, 242, by = 3)
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
104 nb_onto <- round(nrow(profile[[1]]) / 25) * 25
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
105 if (nb_onto < 25) {
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
106 nb_onto <- 25
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
107 }
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
108 if (nb_onto <= 2000) {
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
109 height <- as.integer(names(which(range == nb_onto)))
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
110 } else {
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
111 height <- 250
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
112 }
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
113 return(c(width, height))
9
948fecb6a40b planemo upload commit 973d782455fd6a7df7ba9bce0a7878de53bc1e68-dirty
proteore
parents: 8
diff changeset
114 }
948fecb6a40b planemo upload commit 973d782455fd6a7df7ba9bce0a7878de53bc1e68-dirty
proteore
parents: 8
diff changeset
115
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
116 make_plot <- function(profile, percent, title, onto, plot_opt) {
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
117
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
118 tmp <- plot_size_from_nb_onto(profile)
9
948fecb6a40b planemo upload commit 973d782455fd6a7df7ba9bce0a7878de53bc1e68-dirty
proteore
parents: 8
diff changeset
119 width <- tmp[1]
948fecb6a40b planemo upload commit 973d782455fd6a7df7ba9bce0a7878de53bc1e68-dirty
proteore
parents: 8
diff changeset
120 height <- tmp[2]
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
121
8
386145573c19 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents: 7
diff changeset
122 if (plot_opt == "PDF") {
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
123 file_name <- paste("profile_", onto, ".pdf", collapse = "", sep = "")
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
124 pdf(file_name, width = width, height = height)
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
125 } else if (plot_opt == "JPEG") {
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
126 file_name <- paste("profile_", onto, ".jpeg", collapse = "", sep = "")
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
127 jpeg(file_name, width = width, height = height, units = "in", res = 100)
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
128 } else if (plot_opt == "PNG") {
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
129 file_name <- paste("profile_", onto, ".png", collapse = "", sep = "")
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
130 png(file_name, width = width, height = height, units = "in", res = 100)
0
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
131 }
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
132 plotProfiles(profile, percentage = percent,
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
133 multiplePlots = FALSE, aTitle = title)
8
386145573c19 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents: 7
diff changeset
134 dev.off()
0
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
135 }
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
136
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
137 goprofiles <- function() {
1
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
138 args <- commandArgs(TRUE)
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
139 if (length(args) < 1) {
1
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
140 args <- c("--help")
0
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
141 }
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
142
1
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
143 # Help section
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
144 if ("--help" %in% args) {
1
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
145 cat("Selection and Annotation HPA
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
146 Arguments:
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
147 --input_type: type of input (list of id or filename)
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
148 --input: input
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
149 --ncol: the column number which you would like to apply...
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
150 --header: true/false if your file contains a header
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
151 --id_type: the type of input IDs (UniProt/EntrezID)
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
152 --onto_opt: ontology options
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
153 --plot_opt: plot extension options (PDF/JPEG/PNG)
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
154 --level: 1-3
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
155 --per
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
156 --title: title of the plot
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
157 --duplicate: remove dupliate input IDs (true/false)
5
781072a65600 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents: 4
diff changeset
158 --text_output: text output filename \n
781072a65600 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents: 4
diff changeset
159 --species")
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
160 q(save = "no")
0
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
161 }
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
162
1
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
163 # Parse arguments
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
164 parse_args <- function(x) strsplit(sub("^--", "", x), "=")
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
165 args_df <- as.data.frame(do.call("rbind", parse_args(args)))
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
166 args <- as.list(as.character(args_df$V2))
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
167 names(args) <- args_df$V1
1
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
168
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
169
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
170 id_type <- args$id_type
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
171 input_type <- args$input_type
1
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
172 if (input_type == "text") {
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
173 input <- unlist(strsplit(strsplit(args$input, "[ \t\n]+")[[1]], ";"))
5
781072a65600 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents: 4
diff changeset
174 } else if (input_type == "file") {
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
175 filename <- args$input
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
176 ncol <- args$ncol
1
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
177 # Check ncol
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
178 if (! as.numeric(gsub("c", "", ncol)) %% 1 == 0) {
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
179 stop("Please enter an integer for level")
5
781072a65600 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents: 4
diff changeset
180 } else {
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
181 ncol <- as.numeric(gsub("c", "", ncol))
1
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
182 }
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
183 header <- str2bool(args$header)
1
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
184 # Get file content
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
185 file <- read_file(filename, header)
1
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
186 # Extract Protein IDs list
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
187 input <- unlist(strsplit(as.character(file[, ncol]), ";"))
1
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
188 }
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
189 input <- input [which(!is.na(gsub("NA", NA, input)))]
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
190
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
191 if (! any(check_ids(input, id_type))) {
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
192 stop(paste(id_type,
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
193 "not found in your ids list, please check your IDs in input
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
194 or the selected column of your input file"))
5
781072a65600 planemo upload commit 0ce4c81e6d2f7af8c9b52f6c07e83b0319c2adb1-dirty
proteore
parents: 4
diff changeset
195 }
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
196
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
197 ontoopt <- strsplit(args$onto_opt, ",")[[1]]
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
198 onto_pos <- as.integer(gsub("BP", 3, gsub("MF", 2, gsub("CC", 1, ontoopt))))
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
199 plotopt <- args$plot_opt
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
200 level <- args$level
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
201 per <- as.logical(args$per)
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
202 title <- args$title
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
203 duplicate <- str2bool(args$duplicate)
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
204 text_output <- args$text_output
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
205 species <- args$species
1
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
206
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
207 profiles <- getprofile(input, id_type, level, duplicate, species)
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
208
8
386145573c19 planemo upload commit bdd7e8a1f08c11db2a9f1b6db5535c6d32153b2b
proteore
parents: 7
diff changeset
209 for (index in onto_pos) {
15
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
210 onto <- names(profiles[index])
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
211 profile <- profiles[index]
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
212 make_plot(profile, per, title, onto, plotopt)
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
213 text_output <- paste("goProfiles_", onto, "_",
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
214 title, ".tsv", sep = "", collapse = "")
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
215 profile <- as.data.frame(profile)
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
216 profile <- as.data.frame(apply(profile, c(1, 2),
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
217 function(x) gsub("^$|^ $", NA, x))) #convert "" and " " to NA
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
218 write.table(profile, text_output, sep = "\t",
601027649251 "planemo upload commit 6bc2c32177acd823d5025bc90a562e6e6b2a233a"
proteore
parents: 11
diff changeset
219 row.names = FALSE, quote = FALSE, col.names = T)
1
1236ee08ccb8 planemo upload commit 5774fd6a5a746f36f6bf4671a51a39ea2b978300-dirty
proteore
parents: 0
diff changeset
220 }
0
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
221 }
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
222
d89c09253c8d planemo upload commit abb24d36c776520e73220d11386252d848173697-dirty
proteore
parents:
diff changeset
223 goprofiles()