Mercurial > repos > artbio > mutational_patterns

annotate mutational_patterns.R @ 0:924c527fb379 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/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
author artbio
date Sun, 13 Sep 2020 18:40:29 +0000
parents
children aea952be68cb
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
1 # load packages that are provided in the conda env
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
2 options( show.error.messages=F,
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
3 error = function () { cat( geterrmessage(), file=stderr() ); q( "no", 1, F ) } )
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
4 loc <- Sys.setlocale("LC_MESSAGES", "en_US.UTF-8")
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
5 warnings()
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
6 library(optparse)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
7 library(rjson)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
8 library(MutationalPatterns)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
9 library(ggplot2)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
10
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
11 # Arguments
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
12 option_list = list(
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
13 make_option(
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
14 "--inputs",
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
15 default = NA,
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
16 type = 'character',
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
17 help = "json formatted dictionary of datasets and their paths"
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
18 ),
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
19 make_option(
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
20 "--genome",
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
21 default = NA,
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
22 type = 'character',
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
23 help = "genome name in the BSgenome bioconductor package"
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
24 ),
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
25 make_option(
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
26 "--levels",
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
27 default = NA,
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
28 type = 'character',
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
29 help = "path to the tab separated file describing the levels in function of datasets"
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
30 ),
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
31 make_option(
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
32 "--signum",
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
33 default = 2,
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
34 type = 'integer',
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
35 help = "selects the N most significant signatures in samples to express mutational patterns"
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
36 ),
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
37 make_option(
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
38 "--output",
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
39 default = NA,
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
40 type = 'character',
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
41 help = "path to output dataset"
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
42 )
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
43 )
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
44
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
45 opt = parse_args(OptionParser(option_list = option_list),
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
46 args = commandArgs(trailingOnly = TRUE))
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
47
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
48 json_dict <- opt$inputs
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
49 parser <- newJSONParser()
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
50 parser$addData(json_dict)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
51 fileslist <- parser$getObject()
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
52 vcf_files <- attr(fileslist, "names")
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
53 sample_names <- unname(unlist(fileslist))
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
54 pdf(opt$output, paper = "special", width = 11.69, height = 11.69)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
55 ref_genome <- opt$genome
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
56 library(ref_genome, character.only = TRUE)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
57
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
58 # Load the VCF files into a GRangesList:
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
59 vcfs <- read_vcfs_as_granges(vcf_files, sample_names, ref_genome)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
60 levels_table <- read.delim(opt$levels, header=FALSE, col.names=c("sample_name","level"))
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
61 vcf_table <- data.frame(path=vcf_files, sample_name=sample_names)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
62 metadata_table <- merge(vcf_table, levels_table, by.x=2, by.y=1)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
63 levels <- metadata_table$level
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
64 muts = mutations_from_vcf(vcfs[[1]])
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
65 types = mut_type(vcfs[[1]])
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
66 context = mut_context(vcfs[[1]], ref_genome)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
67 type_context = type_context(vcfs[[1]], ref_genome)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
68 type_occurrences <- mut_type_occurrences(vcfs, ref_genome)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
69 # p1 <- plot_spectrum(type_occurrences)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
70 # p2 <- plot_spectrum(type_occurrences, CT = TRUE)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
71 # p3 <- plot_spectrum(type_occurrences, CT = TRUE, legend = FALSE)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
72 #
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
73 # plot(p2)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
74 # p4 <- plot_spectrum(type_occurrences, by = levels, CT = TRUE, legend = TRUE)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
75 # palette <- c("pink", "orange", "blue", "lightblue", "green", "red", "purple")
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
76 # p5 <- plot_spectrum(type_occurrences, CT=TRUE, legend=TRUE, colors=palette)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
77 #
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
78 # plot(p4)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
79 mut_mat <- mut_matrix(vcf_list = vcfs, ref_genome = ref_genome)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
80 # plot_96_profile(mut_mat[,1:length(as.data.frame(mut_mat))], condensed = TRUE)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
81 mut_mat <- mut_mat + 0.0001
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
82 # library("NMF")
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
83 # estimate <- nmf(mut_mat, rank=2:5, method="brunet", nrun=100, seed=123456)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
84 # plot(estimate)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
85 # nmf_res <- extract_signatures(mut_mat, rank = 4, nrun = 100)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
86 # colnames(nmf_res$signatures) <- c("Signature A", "Signature B", "Signature C", "Signature D")
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
87 # rownames(nmf_res$contribution) <- c("Signature A", "Signature B", "Signature C", "Signature D")
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
88 # plot_96_profile(nmf_res$signatures, condensed = TRUE)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
89 sp_url <- paste("https://cancer.sanger.ac.uk/cancergenome/assets/", "signatures_probabilities.txt", sep = "")
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
90 cancer_signatures = read.table(sp_url, sep = "\t", header = TRUE)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
91 new_order = match(row.names(mut_mat), cancer_signatures$Somatic.Mutation.Type)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
92 cancer_signatures = cancer_signatures[as.vector(new_order),]
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
93 row.names(cancer_signatures) = cancer_signatures$Somatic.Mutation.Type
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
94 cancer_signatures = as.matrix(cancer_signatures[,4:33])
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
95 # plot_96_profile(cancer_signatures, condensed = TRUE, ymax = 0.3)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
96 hclust_cosmic = cluster_signatures(cancer_signatures, method = "average")
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
97 cosmic_order = colnames(cancer_signatures)[hclust_cosmic$order]
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
98 # plot(hclust_cosmic)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
99 cos_sim(mut_mat[,1], cancer_signatures[,1])
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
100 cos_sim_samples_signatures = cos_sim_matrix(mut_mat, cancer_signatures)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
101 plot_cosine_heatmap(cos_sim_samples_signatures, col_order = cosmic_order, cluster_rows = TRUE)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
102 fit_res <- fit_to_signatures(mut_mat, cancer_signatures)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
103 threshold <- tail(sort(unlist(rowSums(fit_res$contribution), use.names = FALSE)), opt$signum)[1]
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
104 select <- which(rowSums(fit_res$contribution) >= threshold) # ensure opt$signum best signatures in samples are retained, the others discarded
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
105 plot_contribution(fit_res$contribution[select,], cancer_signatures[,select], coord_flip = T, mode = "absolute")
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
106 plot_contribution(fit_res$contribution[select,], cancer_signatures[,select], coord_flip = T, mode = "relative")
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
107 plot_contribution_heatmap(fit_res$contribution, cluster_samples = TRUE, method = "complete")
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
108
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
109
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
110 sig5data <- as.data.frame(t(head(fit_res$contribution[select,])))
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
111 colnames(sig5data) <- gsub("nature", "", colnames(sig5data))
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
112 sig5data_percents <- sig5data / (apply(sig5data,1,sum)) * 100
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
113 sig5data_percents$sample <- rownames(sig5data_percents)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
114 library(reshape2)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
115 melted_sig5data_percents <-melt(data=sig5data_percents)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
116 melted_sig5data_percents$label <- sub("Sig.", "", melted_sig5data_percents$variable)
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
117 melted_sig5data_percents$pos <- cumsum(melted_sig5data_percents$value) - melted_sig5data_percents$value/2
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
118 ggplot(melted_sig5data_percents, aes(x="", y=value, group=variable, fill=variable)) +
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
119 geom_bar(width = 1, stat = "identity") +
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
120 geom_text(aes(label = label), position = position_stack(vjust = 0.5), color="black", size=3) +
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
121 coord_polar("y", start=0) + facet_wrap(~ sample) +
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
122 labs(x="", y="Samples", fill = "Signatures (Cosmic_v2,March 2015)") +
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
123 theme(axis.text = element_blank(),
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
124 axis.ticks = element_blank(),
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
125 panel.grid = element_blank())
924c527fb379 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mutational_patterns commit e1f3ca871f13569401f41a5af9d0e281bf372540"
artbio
parents:
diff changeset
126 dev.off()