Mercurial > repos > mvdbeek > r_goseq_1_22_0

annotate goseq.r @ 6:0e9424413ab0 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
author mvdbeek
date Thu, 03 Mar 2016 09:56:51 -0500
parents b79c65c90744
children 9ffae7bc23c2
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
6
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
1 options( show.error.messages=F, error = function () { cat( geterrmessage(), file=stderr() ); q( "no", 1, F ) } )
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
2
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
3 # we need that to not crash galaxy with an UTF8 error on German LC settings.
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
4 loc <- Sys.setlocale("LC_MESSAGES", "en_US.UTF-8")
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
5
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
6 suppressPackageStartupMessages({
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
7 library("goseq")
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
8 library("optparse")
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
9 })
0
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
10
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
11 option_list <- list(
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
12 make_option(c("-d", "--dge_file"), type="character", help="Path to file with differential gene expression result"),
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
13 make_option(c("-w","--wallenius_tab"), type="character", help="Path to output file with P-values estimated using wallenius distribution."),
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
14 make_option(c("-s","--sampling_tab"), type="character", default=FALSE, help="Path to output file with P-values estimated using wallenius distribution."),
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
15 make_option(c("-n","--nobias_tab"), type="character", default=FALSE, help="Path to output file with P-values estimated using wallenius distribution and no correction for gene length bias."),
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
16 make_option(c("-l","--length_bias_plot"), type="character", default=FALSE, help="Path to length-bias plot."),
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
17 make_option(c("-sw","--sample_vs_wallenius_plot"), type="character", default=FALSE, help="Path to plot comparing sampling with wallenius p-values."),
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
18 make_option(c("-padj", "--p_adj_column"), type="integer",help="Column that contains p. adjust values"),
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
19 make_option(c("-c", "--cutoff"), type="double",dest="p_adj_cutoff",
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
20 help="Genes with p.adjust below cutoff are considered not differentially expressed and serve as control genes"),
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
21 make_option(c("-r", "--repcnt"), type="integer", default=100, help="Number of repeats for sampling"),
4
76eab486aba9 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents: 3
diff changeset
22 make_option(c("-lf", "--length_file"), type="character", default="FALSE", help = "Path to tabular file mapping gene id to length"),
6
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
23 make_option(c("-cat_file", "--category_file"), default="FALSE", type="character", help = "Path to tabular file with gene_id <-> category mapping."),
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
24 make_option(c("-g", "--genome"), default=NULL, type="character", help = "Genome [used for looking up correct gene length]"),
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
25 make_option(c("-i", "--gene_id"), default=NULL, type="character", help = "Gene ID format of genes in DGE file"),
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
26 make_option(c("-cat", "--use_genes_without_cat"), default=FALSE, type="logical",
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
27 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)."),
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
28 make_option(c("-plots", "--make_plots"), default=FALSE, type="logical", help="produce diagnostic plots?")
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
29 )
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
30
0
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
31 parser <- OptionParser(usage = "%prog [options] file", option_list=option_list)
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
32 args = parse_args(parser)
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
33
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
34 # Vars:
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
35 dge_file = args$dge_file
6
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
36 category_file = args$category_file
0
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
37 p_adj_column = args$p_adj_colum
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
38 p_adj_cutoff = args$p_adj_cutoff
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
39 length_file = args$length_file
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
40 genome = args$genome
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
41 gene_id = args$gene_id
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
42 wallenius_tab = args$wallenius_tab
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
43 sampling_tab = args$sampling_tab
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
44 nobias_tab = args$nobias_tab
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
45 length_bias_plot = args$length_bias_plot
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
46 sample_vs_wallenius_plot = args$sample_vs_wallenius_plot
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
47 repcnt = args$repcnt
5
b79c65c90744 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 81aedf1b50849160f6c048c0da4bb1038bb813a5
mvdbeek
parents: 4
diff changeset
48 use_genes_without_cat = args$use_genes_without_cat
6
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
49 make_plots = args$make_plots
0
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
50
6
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
51 # format DE genes into named vector suitable for goseq
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
52 first_line = read.delim(dge_file, header = FALSE, nrow=1)
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
53 # check if header [character where numeric is expected]
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
54 if (is.numeric(first_line[,p_adj_column])) {
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
55 dge_table = read.delim(dge_file, header = FALSE, sep="\t")
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
56 } else {
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
57 dge_table = read.delim(dge_file, header = TRUE, sep="\t")
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
58 }
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
59
0
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
60 genes = as.integer(dge_table[,p_adj_column]<p_adj_cutoff)
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
61 names(genes) = dge_table[,1] # Assuming first row contains gene names
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
62
6
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
63 # gene lengths, assuming last column
4
76eab486aba9 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents: 3
diff changeset
64 if (length_file != "FALSE" ) {
6
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
65 first_line = read.delim(dge_file, header = FALSE, nrow=1)
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
66 if (is.numeric(first_line[, ncol(first_line)])) {
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
67 length_table = read.delim(length_file, header=FALSE, sep="\t", check.names=FALSE)
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
68 } else {
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
69 length_table = read.delim(length_file, header=TRUE, sep="\t", check.names=FALSE)
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
70 }
4
76eab486aba9 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents: 3
diff changeset
71 row.names(length_table) = length_table[,1]
6
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
72 gene_lengths = length_table[names(genes),][,ncol(length_table)]
0
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
73 } else {
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
74 gene_lengths = getlength(names(genes), genome, gene_id)
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
75 }
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
76
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
77 # Estimate PWF
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
78
6
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
79 if (make_plots == TRUE) {
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
80 pdf(length_bias_plot)
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
81 }
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
82 pwf=nullp(genes, genome = genome, id = gene_id, bias.data = gene_lengths, plot.fit=make_plots)
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
83 graphics.off()
5
b79c65c90744 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 81aedf1b50849160f6c048c0da4bb1038bb813a5
mvdbeek
parents: 4
diff changeset
84
6
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
85 # Fetch GO annotations if category_file hasn't been supplied:
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
86 if (category_file == "FALSE") {
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
87 go_map=getgo(genes = names(genes), genome = genome, id = gene_id, fetch.cats=c("GO:CC", "GO:BP", "GO:MF", "KEGG"))
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
88 } else {
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
89 # check for header: first entry in first column must be present in genes, else it's a header
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
90 first_line = read.delim(category_file, header = FALSE, nrow=1)
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
91 if (first_line[,1] %in% names(genes)) {
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
92 go_map = read.delim(category_file, header = FALSE)
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
93 } else {
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
94 go_map = read.delim(category_file, header= TRUE)
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
95 }
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
96 }
5
b79c65c90744 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 81aedf1b50849160f6c048c0da4bb1038bb813a5
mvdbeek
parents: 4
diff changeset
97
0
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
98 # wallenius approximation of p-values
6
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
99 GO.wall=goseq(pwf, genome = genome, id = gene_id, use_genes_without_cat = use_genes_without_cat, gene2cat=go_map)
0
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
100
6
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
101 GO.nobias=goseq(pwf, genome = genome, id = gene_id, method="Hypergeometric", use_genes_without_cat = use_genes_without_cat, gene2cat=go_map)
0
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
102
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
103 # Sampling distribution
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
104 if (repcnt > 0) {
6
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
105 GO.samp=goseq(pwf, genome = genome, id = gene_id, method="Sampling", repcnt=repcnt, use_genes_without_cat = use_genes_without_cat, gene2cat=go_map)
0
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
106 # Compare sampling with wallenius
6
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
107 if (make_plots == TRUE) {
0
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
108 pdf(sample_vs_wallenius_plot)
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
109 plot(log10(GO.wall[,2]), log10(GO.samp[match(GO.samp[,1],GO.wall[,1]),2]),
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
110 xlab="log10(Wallenius p-values)",ylab="log10(Sampling p-values)",
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
111 xlim=c(-3,0))
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
112 abline(0,1,col=3,lty=2)
6
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
113 graphics.off()
0e9424413ab0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit f95b47ed1a09ce14d3b565e8ea56d8bf12c35814-dirty
mvdbeek
parents: 5
diff changeset
114 }
0
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
115 write.table(GO.samp, sampling_tab, sep="\t", row.names = FALSE, quote = FALSE)
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
116 }
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
117
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
118
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
119 write.table(GO.wall, wallenius_tab, sep="\t", row.names = FALSE, quote = FALSE)
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
120 write.table(GO.nobias, nobias_tab, sep="\t", row.names = FALSE, quote = FALSE)
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
121
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
122 sessionInfo()
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
123
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
124 # Use the following to get a list of supported genomes / gene ids
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
125
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
126 # write.table(supportedGenomes(), "available_genomes.tab", row.names = FALSE, quote=FALSE)
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
127 # write.table(supportedGeneIDs(), "supported_gene_ids.tab", row.name = FALSE, quote = FALSE)
a6427f7893b0 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
mvdbeek
parents:
diff changeset
128 # write.table(table.summary, "input_gene_count_matrix.tab", row.names = FALSE, quote = FALSE)