Mercurial > repos > iuc > goseq
annotate goseq.r @ 7:53eff3348302 draft
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
| author | iuc | 
|---|---|
| date | Fri, 06 Sep 2019 07:50:23 -0400 | 
| parents | 85f3b5ab2bf8 | 
| children | 7c9731142105 | 
| rev | line source | 
|---|---|
| 
0
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
1 options( show.error.messages=F, error = function () { cat( geterrmessage(), file=stderr() ); q( "no", 1, F ) } ) | 
| 
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
2 | 
| 
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
3 # we need that to not crash galaxy with an UTF8 error on German LC settings. | 
| 
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
4 loc <- Sys.setlocale("LC_MESSAGES", "en_US.UTF-8") | 
| 
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
5 | 
| 
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
6 suppressPackageStartupMessages({ | 
| 
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
7 library("goseq") | 
| 
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
8 library("optparse") | 
| 
2
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
9 library("dplyr") | 
| 
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
10 library("ggplot2") | 
| 
0
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
11 }) | 
| 
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
12 | 
| 
7
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
13 sessionInfo() | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
14 | 
| 
0
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
15 option_list <- list( | 
| 
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
16 make_option(c("-d", "--dge_file"), type="character", help="Path to file with differential gene expression result"), | 
| 
7
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
17 make_option(c("-lf", "--length_file"), type="character", default=NULL, help="Path to tabular file mapping gene id to length"), | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
18 make_option(c("-g", "--genome"), type="character", default=NULL, help="Genome [used for looking up correct gene length]"), | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
19 make_option(c("-i", "--gene_id"), type="character", default=NULL, help="Gene ID format of genes in DGE file"), | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
20 make_option(c("-fc", "--fetch_cats"), type="character", default=NULL, help="Categories to get can include one or more of GO:CC, GO:BP, GO:MF, KEGG"), | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
21 make_option(c("-cat_file", "--category_file"), type="character", default=NULL, help="Path to tabular file with gene_id <-> category mapping"), | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
22 make_option(c("-w","--wallenius_tab"), type="character", default=NULL, help="Path to output file with P-values estimated using wallenius distribution"), | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
23 make_option(c("-n","--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"), | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
24 make_option(c("-r", "--repcnt"), type="integer", default=0, help="Number of repeats for sampling"), | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
25 make_option(c("-s","--sampling_tab"), type="character", default=NULL, help="Path to output file with P-values estimated using sampling distribution"), | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
26 make_option(c("-p", "--p_adj_method"), type="character", default="BH", help="Multiple hypothesis testing correction method to use"), | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
27 make_option(c("-cat", "--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)."), | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
28 make_option(c("-tp", "--top_plot"), type="character", default=NULL, help="Path to output PDF with top10 over-rep GO terms"), | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
29 make_option(c("-plots", "--make_plots"), default=FALSE, type="logical", help="Produce diagnostic plots?"), | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
30 make_option(c("-l","--length_bias_plot"), type="character", default=NULL, help="Path to length-bias plot"), | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
31 make_option(c("-sw","--sample_vs_wallenius_plot"), type="character", default=NULL, help="Path to plot comparing sampling with wallenius p-values"), | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
32 make_option(c("-rd", "--rdata"), type="character", default=NULL, help="Path to RData output file"), | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
33 make_option(c("-g2g", "--categories_genes_out_fp"), type="character", default=NULL, help="Path to file with categories (GO/KEGG terms) and associated DE genes") | 
| 
0
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
34 ) | 
| 
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
35 | 
| 
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
36 parser <- OptionParser(usage = "%prog [options] file", option_list=option_list) | 
| 
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
37 args = parse_args(parser) | 
| 
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
38 | 
| 
2
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
39 if (!is.null(args$fetch_cats)) { | 
| 
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
40 fetch_cats = unlist(strsplit(args$fetch_cats, ",")) | 
| 
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
41 } else { | 
| 
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
42 fetch_cats = "Custom" | 
| 
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
43 } | 
| 
0
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
44 | 
| 
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
45 # format DE genes into named vector suitable for goseq | 
| 
2
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
46 # check if header is present | 
| 
7
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
47 first_line = read.delim(args$dge_file, header=FALSE, nrow=1) | 
| 
2
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
48 second_col = toupper(first_line[, ncol(first_line)]) | 
| 
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
49 if (second_col == TRUE || second_col == FALSE) { | 
| 
7
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
50 dge_table = read.delim(args$dge_file, header=FALSE, sep="\t") | 
| 
2
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
51 } else { | 
| 
7
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
52 dge_table = read.delim(args$dge_file, header=TRUE, sep="\t") | 
| 
2
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
53 } | 
| 
7
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
54 genes = as.numeric(as.logical(dge_table[, ncol(dge_table)])) # Last column contains TRUE/FALSE | 
| 
0
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
55 names(genes) = dge_table[,1] # Assuming first column contains gene names | 
| 
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
56 | 
| 
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
57 # gene lengths, assuming last column | 
| 
7
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
58 first_line = read.delim(args$length_file, header=FALSE, nrow=1) | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
59 if (is.numeric(first_line[, ncol(first_line)])) { | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
60 length_table = read.delim(args$length_file, header=FALSE, sep="\t", check.names=FALSE) | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
61 } else { | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
62 length_table = read.delim(args$length_file, header=TRUE, sep="\t", check.names=FALSE) | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
63 } | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
64 row.names(length_table) = length_table[,1] | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
65 # get vector of gene length in same order as the genes | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
66 gene_lengths = length_table[names(genes),][, ncol(length_table)] | 
| 
0
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
67 | 
| 
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
68 # Estimate PWF | 
| 
7
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
69 if (args$make_plots) { | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
70 pdf(args$length_bias_plot) | 
| 
0
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
71 } | 
| 
7
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
72 pwf=nullp(genes, genome=args$genome, id=args$gene_id, bias.data=gene_lengths, plot.fit=args$make_plots) | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
73 if (args$make_plots) { | 
| 
2
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
74 dev.off() | 
| 
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
75 } | 
| 
0
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
76 | 
| 
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
77 # Fetch GO annotations if category_file hasn't been supplied: | 
| 
7
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
78 if (is.null(args$category_file)) { | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
79 go_map=getgo(genes=names(genes), genome=args$genome, id=args$gene_id, fetch.cats=fetch_cats) | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
80 } else { | 
| 
0
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
81 # check for header: first entry in first column must be present in genes, else it's a header | 
| 
7
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
82 first_line = read.delim(args$category_file, header=FALSE, nrow=1) | 
| 
0
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
83 if (first_line[,1] %in% names(genes)) { | 
| 
7
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
84 go_map = read.delim(args$category_file, header=FALSE) | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
85 } else { | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
86 go_map = read.delim(args$category_file, header=TRUE) | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
87 } | 
| 
0
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
88 } | 
| 
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
89 | 
| 
2
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
90 results <- list() | 
| 
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
91 | 
| 
7
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
92 runGoseq <- function(pwf, genome, gene_id, goseq_method, use_genes_without_cat, repcnt, gene2cat, p_adj_method, out_fp){ | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
93 out=goseq(pwf, genome=genome, id=gene_id, method=goseq_method, use_genes_without_cat=use_genes_without_cat, gene2cat=go_map) | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
94 out$p.adjust.over_represented = p.adjust(out$over_represented_pvalue, method=p_adj_method) | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
95 out$p.adjust.under_represented = p.adjust(out$under_represented_pvalue, method=p_adj_method) | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
96 write.table(out, out_fp, sep="\t", row.names=FALSE, quote=FALSE) | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
97 return(out) | 
| 
0
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
98 } | 
| 
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
99 | 
| 
7
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
100 # wallenius approximation of p-values | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
101 if (!is.null(args$wallenius_tab)) results[['Wallenius']] <- runGoseq( | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
102 pwf, | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
103 genome=args$genome, | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
104 gene_id=args$gene_id, | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
105 goseq_method="Wallenius", | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
106 use_genes_without_cat=args$use_genes_without_cat, | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
107 repcnt=args$repcnt, | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
108 gene2cat=go_map, | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
109 p_adj_method=args$p_adj_method, | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
110 out_fp=args$wallenius_tab) | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
111 | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
112 | 
| 
0
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
113 # hypergeometric (no length bias correction) | 
| 
7
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
114 if (!is.null(args$nobias_tab)) results[['Hypergeometric']] <- runGoseq( | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
115 pwf, | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
116 genome=args$genome, | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
117 gene_id=args$gene_id, | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
118 goseq_method="Hypergeometric", | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
119 use_genes_without_cat=args$use_genes_without_cat, | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
120 repcnt=args$repcnt, | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
121 gene2cat=go_map, | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
122 p_adj_method=args$p_adj_method, | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
123 out_fp=args$nobias_tab) | 
| 
0
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
124 | 
| 
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
125 # Sampling distribution | 
| 
7
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
126 if (args$repcnt > 0){ | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
127 results[['Sampling']] <- runGoseq( | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
128 pwf, | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
129 genome=args$genome, | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
130 gene_id=args$gene_id, | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
131 goseq_method="Sampling", | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
132 use_genes_without_cat=args$use_genes_without_cat, | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
133 repcnt=args$repcnt, | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
134 gene2cat=go_map, | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
135 p_adj_method=args$p_adj_method, | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
136 out_fp=args$sampling_tab) | 
| 
4
 
9e6226b6d2ad
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 2dffcdff171ebf695ef0fc7eb8f8264f0d132115
 
iuc 
parents: 
3 
diff
changeset
 | 
137 | 
| 
0
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
138 # Compare sampling with wallenius | 
| 
7
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
139 if (args$make_plots & !is.null(args$wallenius_tab)) { | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
140 pdf(args$sample_vs_wallenius_plot) | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
141 plot(log10(results[['Wallenius']][,2]), | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
142 log10(results[['Sampling']][match(results[['Sampling']][,1], results[['Wallenius']][,1]), 2]), | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
143 xlab="log10(Wallenius p-values)", | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
144 ylab="log10(Sampling p-values)", | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
145 xlim=c(-3,0)) | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
146 abline(0,1,col=3,lty=2) | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
147 dev.off() | 
| 
0
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
148 } | 
| 
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
149 } | 
| 
 
084d77515343
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit b7dcd020c6a15fa55f392cc09cbc37580d6e75c4
 
iuc 
parents:  
diff
changeset
 | 
150 | 
| 
7
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
151 # Plot the top 10 | 
| 
2
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
152 if (!is.null(args$top_plot)) { | 
| 
3
 
d93aa41a7522
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 0798278a90c08228a386516881680b328fc33f0c
 
iuc 
parents: 
2 
diff
changeset
 | 
153 cats_title <- gsub("GO:","", args$fetch_cats) | 
| 
2
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
154 # modified from https://bioinformatics-core-shared-training.github.io/cruk-summer-school-2018/RNASeq2018/html/06_Gene_set_testing.nb.html | 
| 
7
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
155 pdf(args$top_plot) | 
| 
2
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
156 for (m in names(results)) { | 
| 
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
157 p <- results[[m]] %>% | 
| 
4
 
9e6226b6d2ad
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 2dffcdff171ebf695ef0fc7eb8f8264f0d132115
 
iuc 
parents: 
3 
diff
changeset
 | 
158 top_n(10, wt=-over_represented_pvalue) %>% | 
| 
2
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
159 mutate(hitsPerc=numDEInCat*100/numInCat) %>% | 
| 
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
160 ggplot(aes(x=hitsPerc, | 
| 
6
 
85f3b5ab2bf8
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e60c1cecc4561d61624799b0999e89687677ab1
 
iuc 
parents: 
4 
diff
changeset
 | 
161 y=reorder(substr(term, 1, 40), -over_represented_pvalue), # only use 1st 40 chars of terms otherwise squashes plot | 
| 
 
85f3b5ab2bf8
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e60c1cecc4561d61624799b0999e89687677ab1
 
iuc 
parents: 
4 
diff
changeset
 | 
162 colour=p.adjust.over_represented, | 
| 
2
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
163 size=numDEInCat)) + | 
| 
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
164 geom_point() + | 
| 
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
165 expand_limits(x=0) + | 
| 
6
 
85f3b5ab2bf8
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e60c1cecc4561d61624799b0999e89687677ab1
 
iuc 
parents: 
4 
diff
changeset
 | 
166 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")) + | 
| 
7
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
167 theme(plot.title=element_text(hjust = 0.5), plot.subtitle=element_text(hjust = 0.5)) | 
| 
2
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
168 print(p) | 
| 
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
169 } | 
| 
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
170 dev.off() | 
| 
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
171 } | 
| 
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
172 | 
| 
7
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
173 # Extract the genes to the categories (GO/KEGG terms) | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
174 if (!is.null(args$categories_genes_out_fp)) { | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
175 cat2gene = split(rep(names(go_map), sapply(go_map, length)), unlist(go_map, use.names = FALSE)) | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
176 # extract categories (GO/KEGG terms) for all results | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
177 categories = c() | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
178 for (m in names(results)) { | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
179 categories = c(categories, results[[m]]$category) | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
180 } | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
181 categories = unique(categories) | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
182 # extract the DE genes for each catge term | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
183 categories_genes = data.frame(Categories=categories, DEgenes=rep('', length(categories))) | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
184 categories_genes$DEgenes = as.character(categories_genes$DEgenes) | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
185 rownames(categories_genes) = categories | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
186 for (cat in categories){ | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
187 tmp = pwf[cat2gene[[cat]],] | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
188 tmp = rownames(tmp[tmp$DEgenes > 0, ]) | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
189 categories_genes[cat, 'DEgenes'] = paste(tmp, collapse=',') | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
190 } | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
191 # output | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
192 write.table(categories_genes, args$categories_genes_out_fp, sep = "\t", row.names=FALSE, quote=FALSE) | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
193 } | 
| 
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
194 | 
| 
2
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
195 # Output RData file | 
| 
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
196 if (!is.null(args$rdata)) { | 
| 
7
 
53eff3348302
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 8e19f8bcaea6f607a1eaa14bb88f2d625ed63df0"
 
iuc 
parents: 
6 
diff
changeset
 | 
197 save.image(file=args$rdata) | 
| 
2
 
25e1e7b40e55
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/goseq commit 46c4278d292ab4d76dc5f3f74c3109c3179be7ef
 
iuc 
parents: 
0 
diff
changeset
 | 
198 } | 
