annotate segmentation_sequenza.R @ 6:88e03bac1e36 draft

"planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit ccbc1fc0e1af1e9cf5000fe2a3f60655cd5793eb"
author artbio
date Sun, 07 Mar 2021 12:01:21 +0000
parents e299c477b475
children e66b91475343
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
1 # load packages that are provided in the conda env
6
88e03bac1e36 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit ccbc1fc0e1af1e9cf5000fe2a3f60655cd5793eb"
artbio
parents: 4
diff changeset
2 # options(show.error.messages = F,
88e03bac1e36 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit ccbc1fc0e1af1e9cf5000fe2a3f60655cd5793eb"
artbio
parents: 4
diff changeset
3 # error = function() {
88e03bac1e36 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit ccbc1fc0e1af1e9cf5000fe2a3f60655cd5793eb"
artbio
parents: 4
diff changeset
4 # cat(geterrmessage(), file = stderr()); q("no", 1, F)})
0
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
5
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
6 library(optparse)
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
7 library(sequenza)
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
8 library(BiocParallel)
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
9 library(tidyverse)
6
88e03bac1e36 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit ccbc1fc0e1af1e9cf5000fe2a3f60655cd5793eb"
artbio
parents: 4
diff changeset
10 Sys.setenv(TZ = "Etc/UTC") # turnaround the tidyverse bug "In OlsonNames() : no Olson database found"
88e03bac1e36 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit ccbc1fc0e1af1e9cf5000fe2a3f60655cd5793eb"
artbio
parents: 4
diff changeset
11 tzdirs <- c(Sys.getenv("TZDIR"), file.path(R.home("share"),
88e03bac1e36 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit ccbc1fc0e1af1e9cf5000fe2a3f60655cd5793eb"
artbio
parents: 4
diff changeset
12 "zoneinfo"), "/usr/share/zoneinfo", "/usr/share/lib/zoneinfo",
88e03bac1e36 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit ccbc1fc0e1af1e9cf5000fe2a3f60655cd5793eb"
artbio
parents: 4
diff changeset
13 "/usr/lib/zoneinfo", "/usr/local/etc/zoneinfo", "/etc/zoneinfo",
88e03bac1e36 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit ccbc1fc0e1af1e9cf5000fe2a3f60655cd5793eb"
artbio
parents: 4
diff changeset
14 "/usr/etc/zoneinfo")
0
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
15 option_list <- list(
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
16 make_option(
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
17 c("-i", "--input"),
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
18 default = NA,
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
19 type = "character",
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
20 help = "Path to Sequenza seqz processed segments file"
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
21 ),
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
22 make_option(
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
23 c("-O", "--output_dir"),
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
24 default = NA,
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
25 type = "character",
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
26 help = "Output directory"
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
27 ),
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
28 make_option(
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
29 c("-s", "--sample_name"),
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
30 default = NA,
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
31 type = "character",
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
32 help = "Sample name"
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
33 )
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
34 )
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
35
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
36 opt <- parse_args(OptionParser(option_list = option_list),
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
37 args = commandArgs(trailingOnly = TRUE))
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
38
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
39 data_file <- opt$input
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
40 output_dir <- opt$output_dir
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
41 sample_name <- opt$sample_name
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
42
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
43
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
44 ## Processing seqz files : normalisation and segmentation for chromosomes 1 to 22
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
45 message(sprintf("\nExtraction step for %s", data_file))
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
46
4
e299c477b475 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit 10edea05400e21ba43942bd0cc582153d60ae2d9"
artbio
parents: 2
diff changeset
47 chrom_list <- c("chr1", "chr2", "chr3", "chr4", "chr5", "chr6", "chr7", "chr8", "chr9",
2
9265faa93098 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8-dirty"
artbio
parents: 0
diff changeset
48 "chr10", "chr11", "chr12", "chr13", "chr14", "chr15", "chr16", "chr17",
9265faa93098 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8-dirty"
artbio
parents: 0
diff changeset
49 "chr18", "chr19", "chr20", "chr21", "chr22")
9265faa93098 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8-dirty"
artbio
parents: 0
diff changeset
50
0
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
51 segfile <- sequenza.extract(data_file,
2
9265faa93098 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8-dirty"
artbio
parents: 0
diff changeset
52 verbose = TRUE,
9265faa93098 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8-dirty"
artbio
parents: 0
diff changeset
53 chromosome.list = chrom_list)
0
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
54
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
55 ## Estimation of cellularity and ploidy
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
56
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
57 segfile_cp <- sequenza.fit(segfile)
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
58 message(sprintf("\nEstimation step for %s\n", data_file))
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
59
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
60 ## writing files and plots using default parameters
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
61 message(sprintf("\nWriting files and plots for %s\n", data_file))
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
62
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
63 sequenza.results(sequenza.extract = segfile,
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
64 cp.table = segfile_cp,
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
65 sample.id = sample_name,
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
66 out.dir = output_dir)
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
67 message(sprintf("\nOutput written to %s\n", output_dir))