annotate goseq.r @ 11:602de62d995b draft default tip

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 0829a85b7fd9d09202450e50d30446cf9845791c
author iuc
date Wed, 11 Jan 2023 10:13:15 +0000
parents ef2ad746b589
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
11
602de62d995b planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 0829a85b7fd9d09202450e50d30446cf9845791c
iuc
parents: 9
diff changeset
1 options(show.error.messages = FALSE, error = function() {
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
2 cat(geterrmessage(), file = stderr())
11
602de62d995b planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 0829a85b7fd9d09202450e50d30446cf9845791c
iuc
parents: 9
diff changeset
3 q("no", 1, FALSE)
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
4 })
0
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
5
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
6 # we need that to not crash galaxy with an UTF8 error on German LC settings.
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
7 loc <- Sys.setlocale("LC_MESSAGES", "en_US.UTF-8")
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
8
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
9 suppressPackageStartupMessages({
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
10 library("goseq")
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
11 library("optparse")
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
12 library("dplyr")
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
13 library("ggplot2")
0
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
14 })
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
15
8
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
16 sessionInfo()
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
17
0
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
18 option_list <- list(
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
19 make_option("--dge_file", type = "character", help = "Path to file with differential gene expression result"),
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
20 make_option("--length_file", type = "character", default = NULL, help = "Path to tabular file mapping gene id to length"),
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
21 make_option("--genome", type = "character", default = NULL, help = "Genome [used for looking up correct gene length]"),
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
22 make_option("--gene_id", type = "character", default = NULL, help = "Gene ID format of genes in DGE file"),
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
23 make_option("--fetch_cats", type = "character", default = NULL, help = "Categories to get can include one or more of GO:CC, GO:BP, GO:MF, KEGG"),
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
24 make_option("--category_file", type = "character", default = NULL, help = "Path to tabular file with gene_id <-> category mapping"),
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
25 make_option("--wallenius_tab", type = "character", default = NULL, help = "Path to output file with P-values estimated using wallenius distribution"),
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
26 make_option("--nobias_tab", type = "character", default = NULL, help = "Path to output file with P-values estimated using hypergeometric distribution and no correction for gene length bias"),
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
27 make_option("--repcnt", type = "integer", default = 0, help = "Number of repeats for sampling"),
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
28 make_option("--sampling_tab", type = "character", default = NULL, help = "Path to output file with P-values estimated using sampling distribution"),
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
29 make_option("--p_adj_method", type = "character", default = "BH", help = "Multiple hypothesis testing correction method to use"),
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
30 make_option("--use_genes_without_cat", type = "logical", default = FALSE, help = "A large number of gene may have no GO term annotated. If this option is set to FALSE, genes without category will be ignored in the calculation of p-values(default behaviour). If TRUE these genes will count towards the total number of genes outside the tested category (default behaviour prior to version 1.15.2)."),
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
31 make_option("--top_plot", type = "character", default = NULL, help = "Path to output PDF with top10 over-rep GO terms"),
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
32 make_option("--make_plots", default = FALSE, type = "logical", help = "Produce diagnostic plots?"),
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
33 make_option("--length_bias_plot", type = "character", default = NULL, help = "Path to length-bias plot"),
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
34 make_option("--sample_vs_wallenius_plot", type = "character", default = NULL, help = "Path to plot comparing sampling with wallenius p-values"),
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
35 make_option("--rdata", type = "character", default = NULL, help = "Path to RData output file"),
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
36 make_option("--categories_genes_out_fp", type = "character", default = NULL, help = "Path to file with categories (GO/KEGG terms) and associated DE genes")
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
37 )
0
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
38
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
39 parser <- OptionParser(usage = "%prog [options] file", option_list = option_list)
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
40 args <- parse_args(parser)
0
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
41
2
ab492df30cdf planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 4a3c9f195ba5d899b1a1ce5e80281cdf230f456a
iuc
parents: 0
diff changeset
42 if (!is.null(args$fetch_cats)) {
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
43 fetch_cats <- unlist(strsplit(args$fetch_cats, ","))
3
783e8b70b047 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
iuc
parents: 2
diff changeset
44 } else {
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
45 fetch_cats <- "Custom"
2
ab492df30cdf planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 4a3c9f195ba5d899b1a1ce5e80281cdf230f456a
iuc
parents: 0
diff changeset
46 }
0
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
47
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
48 # format DE genes into named vector suitable for goseq
2
ab492df30cdf planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 4a3c9f195ba5d899b1a1ce5e80281cdf230f456a
iuc
parents: 0
diff changeset
49 # check if header is present
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
50 first_line <- read.delim(args$dge_file, header = FALSE, nrow = 1)
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
51 second_col <- toupper(first_line[, ncol(first_line)])
2
ab492df30cdf planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 4a3c9f195ba5d899b1a1ce5e80281cdf230f456a
iuc
parents: 0
diff changeset
52 if (second_col == TRUE || second_col == FALSE) {
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
53 dge_table <- read.delim(args$dge_file, header = FALSE, sep = "\t")
2
ab492df30cdf planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 4a3c9f195ba5d899b1a1ce5e80281cdf230f456a
iuc
parents: 0
diff changeset
54 } else {
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
55 dge_table <- read.delim(args$dge_file, header = TRUE, sep = "\t")
2
ab492df30cdf planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 4a3c9f195ba5d899b1a1ce5e80281cdf230f456a
iuc
parents: 0
diff changeset
56 }
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
57 genes <- as.numeric(as.logical(dge_table[, ncol(dge_table)])) # Last column contains TRUE/FALSE
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
58 names(genes) <- dge_table[, 1] # Assuming first column contains gene names
0
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
59
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
60 # gene lengths, assuming last column
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
61 first_line <- read.delim(args$length_file, header = FALSE, nrow = 1)
8
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
62 if (is.numeric(first_line[, ncol(first_line)])) {
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
63 length_table <- read.delim(args$length_file, header = FALSE, sep = "\t", check.names = FALSE)
8
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
64 } else {
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
65 length_table <- read.delim(args$length_file, header = TRUE, sep = "\t", check.names = FALSE)
8
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
66 }
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
67 row.names(length_table) <- length_table[, 1]
8
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
68 # get vector of gene length in same order as the genes
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
69 gene_lengths <- length_table[names(genes), ][, ncol(length_table)]
0
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
70
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
71 # Estimate PWF
8
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
72 if (args$make_plots) {
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
73 pdf(args$length_bias_plot)
0
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
74 }
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
75 pwf <- nullp(genes, genome = args$genome, id = args$gene_id, bias.data = gene_lengths, plot.fit = args$make_plots)
8
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
76 if (args$make_plots) {
3
783e8b70b047 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
iuc
parents: 2
diff changeset
77 dev.off()
783e8b70b047 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
iuc
parents: 2
diff changeset
78 }
0
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
79
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
80 # Fetch GO annotations if category_file hasn't been supplied:
8
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
81 if (is.null(args$category_file)) {
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
82 go_map <- getgo(genes = names(genes), genome = args$genome, id = args$gene_id, fetch.cats = fetch_cats)
8
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
83 } else {
0
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
84 # check for header: first entry in first column must be present in genes, else it's a header
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
85 first_line <- read.delim(args$category_file, header = FALSE, nrow = 1)
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
86 if (first_line[, 1] %in% names(genes)) {
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
87 go_map <- read.delim(args$category_file, header = FALSE)
8
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
88 } else {
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
89 go_map <- read.delim(args$category_file, header = TRUE)
8
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
90 }
0
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
91 }
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
92
3
783e8b70b047 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
iuc
parents: 2
diff changeset
93 results <- list()
783e8b70b047 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
iuc
parents: 2
diff changeset
94
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
95 run_goseq <- function(pwf, genome, gene_id, goseq_method, use_genes_without_cat, repcnt, gene2cat, p_adj_method, out_fp) {
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
96 out <- goseq(pwf, genome = genome, id = gene_id, method = goseq_method, use_genes_without_cat = use_genes_without_cat, gene2cat = go_map)
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
97 out$p_adjust_over_represented <- p.adjust(out$over_represented_pvalue, method = p_adj_method)
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
98 out$p_adjust_under_represented <- p.adjust(out$under_represented_pvalue, method = p_adj_method)
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
99 write.table(out, out_fp, sep = "\t", row.names = FALSE, quote = FALSE)
8
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
100 return(out)
0
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
101 }
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
102
8
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
103 # wallenius approximation of p-values
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
104 if (!is.null(args$wallenius_tab)) {
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
105 results[["Wallenius"]] <- run_goseq(
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
106 pwf,
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
107 genome = args$genome,
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
108 gene_id = args$gene_id,
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
109 goseq_method = "Wallenius",
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
110 use_genes_without_cat = args$use_genes_without_cat,
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
111 repcnt = args$repcnt,
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
112 gene2cat = go_map,
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
113 p_adj_method = args$p_adj_method,
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
114 out_fp = args$wallenius_tab
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
115 )
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
116 }
8
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
117
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
118
0
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
119 # hypergeometric (no length bias correction)
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
120 if (!is.null(args$nobias_tab)) {
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
121 results[["Hypergeometric"]] <- run_goseq(
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
122 pwf,
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
123 genome = args$genome,
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
124 gene_id = args$gene_id,
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
125 goseq_method = "Hypergeometric",
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
126 use_genes_without_cat = args$use_genes_without_cat,
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
127 repcnt = args$repcnt,
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
128 gene2cat = go_map,
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
129 p_adj_method = args$p_adj_method,
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
130 out_fp = args$nobias_tab
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
131 )
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
132 }
0
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
133
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
134 # Sampling distribution
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
135 if (args$repcnt > 0) {
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
136 results[["Sampling"]] <- run_goseq(
8
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
137 pwf,
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
138 genome = args$genome,
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
139 gene_id = args$gene_id,
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
140 goseq_method = "Sampling",
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
141 use_genes_without_cat = args$use_genes_without_cat,
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
142 repcnt = args$repcnt,
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
143 gene2cat = go_map,
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
144 p_adj_method = args$p_adj_method,
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
145 out_fp = args$sampling_tab
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
146 )
5
bbcf5f7f2af2 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 2dffcdff171ebf695ef0fc7eb8f8264f0d132115
iuc
parents: 4
diff changeset
147
0
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
148 # Compare sampling with wallenius
11
602de62d995b planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 0829a85b7fd9d09202450e50d30446cf9845791c
iuc
parents: 9
diff changeset
149 if (args$make_plots && !is.null(args$wallenius_tab)) {
8
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
150 pdf(args$sample_vs_wallenius_plot)
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
151 plot(log10(results[["Wallenius"]][, 2]),
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
152 log10(results[["Sampling"]][match(results[["Sampling"]][, 1], results[["Wallenius"]][, 1]), 2]),
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
153 xlab = "log10(Wallenius p-values)",
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
154 ylab = "log10(Sampling p-values)",
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
155 xlim = c(-3, 0)
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
156 )
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
157 abline(0, 1, col = 3, lty = 2)
8
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
158 dev.off()
0
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
159 }
3
783e8b70b047 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
iuc
parents: 2
diff changeset
160 }
783e8b70b047 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
iuc
parents: 2
diff changeset
161
8
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
162 # Plot the top 10
3
783e8b70b047 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
iuc
parents: 2
diff changeset
163 if (!is.null(args$top_plot)) {
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
164 cats_title <- gsub("GO:", "", args$fetch_cats)
3
783e8b70b047 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
iuc
parents: 2
diff changeset
165 # modified from https://bioinformatics-core-shared-training.github.io/cruk-summer-school-2018/RNASeq2018/html/06_Gene_set_testing.nb.html
8
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
166 pdf(args$top_plot)
3
783e8b70b047 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
iuc
parents: 2
diff changeset
167 for (m in names(results)) {
783e8b70b047 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
iuc
parents: 2
diff changeset
168 p <- results[[m]] %>%
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
169 top_n(10, wt = -over_represented_pvalue) %>%
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
170 mutate(hitsPerc = numDEInCat * 100 / numInCat) %>%
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
171 ggplot(aes(
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
172 x = hitsPerc,
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
173 y = reorder(substr(term, 1, 40), -over_represented_pvalue), # only use 1st 40 chars of terms otherwise squashes plot
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
174 colour = p_adjust_over_represented,
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
175 size = numDEInCat
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
176 )) +
3
783e8b70b047 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
iuc
parents: 2
diff changeset
177 geom_point() +
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
178 expand_limits(x = 0) +
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
179 labs(x = "% DE in category", y = "Category", colour = "Adj P value", size = "Count", title = paste("Top over-represented categories in", cats_title), subtitle = paste(m, " method")) +
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
180 theme(plot.title = element_text(hjust = 0.5), plot.subtitle = element_text(hjust = 0.5))
3
783e8b70b047 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
iuc
parents: 2
diff changeset
181 print(p)
783e8b70b047 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
iuc
parents: 2
diff changeset
182 }
783e8b70b047 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
iuc
parents: 2
diff changeset
183 dev.off()
0
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
184 }
ade933eff007 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
iuc
parents:
diff changeset
185
8
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
186 # Extract the genes to the categories (GO/KEGG terms)
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
187 if (!is.null(args$categories_genes_out_fp)) {
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
188 cat2gene <- split(rep(names(go_map), sapply(go_map, length)), unlist(go_map, use.names = FALSE))
8
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
189 # extract categories (GO/KEGG terms) for all results
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
190 categories <- c()
8
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
191 for (m in names(results)) {
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
192 categories <- c(categories, results[[m]]$category)
8
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
193 }
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
194 categories <- unique(categories)
8
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
195 # extract the DE genes for each catge term
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
196 categories_genes <- data.frame(category = categories, de_genes = rep("", length(categories)))
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
197 categories_genes$de_genes <- as.character(categories_genes$de_genes)
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
198 rownames(categories_genes) <- categories
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
199 for (cat in categories) {
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
200 tmp <- pwf[cat2gene[[cat]], ]
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
201 tmp <- rownames(tmp[tmp$DEgenes > 0, ])
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
202 categories_genes[cat, "de_genes"] <- paste(tmp, collapse = ",")
8
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
203 }
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
204 # output
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
205 write.table(categories_genes, args$categories_genes_out_fp, sep = "\t", row.names = FALSE, quote = FALSE)
8
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
206 }
8b3e3657034e "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
iuc
parents: 7
diff changeset
207
2
ab492df30cdf planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 4a3c9f195ba5d899b1a1ce5e80281cdf230f456a
iuc
parents: 0
diff changeset
208 # Output RData file
ab492df30cdf planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 4a3c9f195ba5d899b1a1ce5e80281cdf230f456a
iuc
parents: 0
diff changeset
209 if (!is.null(args$rdata)) {
9
ef2ad746b589 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit e230a8db9e090c6e0ea9577863ec6153df79e145"
iuc
parents: 8
diff changeset
210 save.image(file = args$rdata)
2
ab492df30cdf planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 4a3c9f195ba5d899b1a1ce5e80281cdf230f456a
iuc
parents: 0
diff changeset
211 }