Mercurial > repos > mvdbeek > r_goseq_1_22_0

annotate goseq.r @ 12:02e88556ce1d 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 fdd0811efc61c31f88ff17096fbe8ee8cfacd766-dirty
author mvdbeek
date Thu, 25 Feb 2016 06:44:44 -0500
parents fe71b97cc1a5
children 8ce951313688
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
1 sink(stdout(), type = "message")
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
2 library(goseq)
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
3 library(optparse)
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
4
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
5 option_list <- list(
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
6 make_option(c("-d", "--dge_file"), type="character", help="Path to file with differential gene expression result"),
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
7 make_option(c("-w","--wallenius_tab"), type="character", help="Path to output file with P-values estimated using wallenius distribution."),
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
8 make_option(c("-s","--sampling_tab"), type="character", default=FALSE, help="Path to output file with P-values estimated using wallenius distribution."),
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
9 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."),
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
10 make_option(c("-l","--length_bias_plot"), type="character", default=FALSE, help="Path to length-bias plot."),
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
11 make_option(c("-sw","--sample_vs_wallenius_plot"), type="character", default=FALSE, help="Path to plot comparing sampling with wallenius p-values."),
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
12 make_option(c("-padj", "--p_adj_column"), type="integer",help="Column that contains p. adjust values"),
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
13 make_option(c("-c", "--cutoff"), type="double",dest="p_adj_cutoff",
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
14 help="Genes with p.adjust below cutoff are considered not differentially expressed and serve as control genes"),
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
15 make_option(c("-r", "--repcnt"), type="integer", default=100, help="Number of repeats for sampling"),
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
16 make_option(c("-g", "--genome"), type="character", help = "Genome [used for looking up correct gene length]"),
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
17 make_option(c("-i", "--gene_id"), type="character", help="Gene ID of gene column in DGE file")
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
18 )
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
19 parser <- OptionParser(usage = "%prog [options] file", option_list=option_list)
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
20 args = parse_args(parser)
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
21
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
22 # Vars:
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
23 dge_file = args$dge_file
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
24 p_adj_column = args$p_adj_colum
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
25 p_adj_cutoff = args$p_adj_cutoff
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
26 genome = args$genome
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
27 gene_id = args$gene_id
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
28 wallenius_tab = args$wallenius_tab
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
29 sampling_tab = args$sampling_tab
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
30 nobias_tab = args$nobias_tab
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
31 length_bias_plot = args$length_bias_plot
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
32 sample_vs_wallenius_plot = args$sample_vs_wallenius_plot
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
33 repcnt = args$repcnt
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
34
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
35
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
36 # format DE genes into vector suitable for use with goseq
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
37 dge_table = read.delim(dge_file, header = TRUE, sep="\t", check.names = FALSE)
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
38 genes = as.integer(dge_table[,p_adj_column]<p_adj_cutoff)
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
39 names(genes) = dge_table[,1] # Assuming first row contains gene names
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
40
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
41 # Estimate PWF
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
42
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
43 pdf(length_bias_plot)
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
44 pwf=nullp(genes, genome , gene_id)
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
45 dev.off()
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
46 # Null dstribution wallenius
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
47 GO.wall=goseq(pwf, genome, gene_id)
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
48
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
49 GO.nobias=goseq(pwf, genome, gene_id, method="Hypergeometric")
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
50
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
51 # Sampling dsitribution
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
52 GO.samp=goseq(pwf,genome, gene_id, method="Sampling",repcnt=repcnt)
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
53
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
54 # Compare sampling with wallenius
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
55 pdf(sample_vs_wallenius_plot)
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
56 plot(log10(GO.wall[,2]), log10(GO.samp[match(GO.samp[,1],GO.wall[,1]),2]),
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
57 xlab="log10(Wallenius p-values)",ylab="log10(Sampling p-values)",
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
58 xlim=c(-3,0))
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
59 abline(0,1,col=3,lty=2)
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
60 dev.off()
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
61
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
62
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
63 write.table(GO.wall, wallenius_tab, sep="\t", row.names = FALSE, quote = FALSE)
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
64 write.table(GO.samp, sampling_tab, sep="\t", row.names = FALSE, quote = FALSE)
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
65 write.table(GO.nobias, nobias_tab, sep="\t", row.names = FALSE, quote = FALSE)
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
66
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
67 # Use the following to get a list of supported genomes / gene ids
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
68
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
69 # write.table(supportedGenomes(), "available_genomes.tab", row.names = FALSE, quote=FALSE)
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
70 # write.table(supportedGeneIDs(), "supported_gene_ids.tab", row.name = FALSE, quote = FALSE)
fe71b97cc1a5 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/goseq_1_22_0 commit 85e0f34aec3045f48ca9b1a00db02de199a99c95-dirty
mvdbeek
parents:
diff changeset
71 # write.table(table.summary, "input_gene_count_matrix.tab", row.names = FALSE, quote = FALSE)