annotate segmentation_sequenza.R @ 3:15b26e44b18d draft

"planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8-dirty"
author artbio
date Fri, 19 Feb 2021 15:22:05 +0000
parents 9265faa93098
children e299c477b475
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
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
2 options(show.error.messages = F,
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
3 error = function() {
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
4 cat(geterrmessage(), file = stderr()); q("no", 1, F)})
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
5 Sys.setenv(TZ = "Pacific/Auckland") # turnaround the tidyverse bug "In OlsonNames() : no Olson database found"
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
6
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
7
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
8 library(optparse)
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
9 library(sequenza)
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
10 library(BiocParallel)
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
11 library(tidyverse)
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
12
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
13 option_list <- list(
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
14 make_option(
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
15 c("-i", "--input"),
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
16 default = NA,
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
17 type = "character",
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
18 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
19 ),
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
20 make_option(
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
21 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
22 default = NA,
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
23 type = "character",
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
24 help = "Output directory"
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
25 ),
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
26 make_option(
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
27 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
28 default = NA,
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
29 type = "character",
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
30 help = "Sample name"
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
31 )
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
32 )
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 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
35 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
36
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
37 data_file <- opt$input
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
38 output_dir <- opt$output_dir
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
39 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
40
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
41
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
42 ## 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
43 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
44
2
9265faa93098 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8-dirty"
artbio
parents: 0
diff changeset
45 chrom_list = c("chr1", "chr2", "chr3", "chr4", "chr5", "chr6", "chr7", "chr8", "chr9",
9265faa93098 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8-dirty"
artbio
parents: 0
diff changeset
46 "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
47 "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
48
0
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
49 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
50 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
51 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
52
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
53 ## 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
54
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
55 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
56 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
57
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
58 ## 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
59 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
60
0e54da14f831 "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/snvtocnv commit fd049d0796cbfaa6c41a7d80e84d3a734b30acc8"
artbio
parents:
diff changeset
61 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
62 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
63 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
64 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
65 message(sprintf("\nOutput written to %s\n", output_dir))