Mercurial > repos > greg > ideas_preprocessor

comparison ideas_preprocessor.R @ 9:0cf23f33d13d draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author greg
date Wed, 24 Jan 2018 10:16:11 -0500
parents 92bb33490bb4
children f45e461b0557
comparison
equal deleted inserted replaced
8:860b59cae5df 9:0cf23f33d13d
1 (.venv) [galaxy@IDEAS ideas_preprocessor]$ cat ideas_preprocessor.R
1 #!/usr/bin/env Rscript 2 #!/usr/bin/env Rscript
2 3
3 suppressPackageStartupMessages(library("data.table")) 4 suppressPackageStartupMessages(library("data.table"))
4 suppressPackageStartupMessages(library("optparse")) 5 suppressPackageStartupMessages(library("optparse"))
5 6
6 option_list <- list( 7 option_list <- list(
7 make_option(c("--chrom_bed_input"), action="store", dest="chrom_bed_input", defaul=NULL, help="Chromosome windows positions file"), 8 make_option(c("--bed_input"), action="store", dest="bed_input", defaul=NULL, help="Chromosome windows positions file"),
8 make_option(c("--exclude_input"), action="store", dest="exclude_input", defaul=NULL, help="File(s) containing regions to exclude"), 9 make_option(c("--exclude_input"), action="store", dest="exclude_input", defaul=NULL, help="File(s) containing regions to exclude"),
9 make_option(c("--bychr"), action="store_true", dest="bychr", defaul=FALSE, help="Separate files by chromosome"), 10 make_option(c("--bychr"), action="store_true", dest="bychr", defaul=FALSE, help="Separate files by chromosome"),
10 make_option(c("--chrom_len_file"), action="store", dest="chrom_len_file", default=NULL, help="Chromosome lengths file"), 11 make_option(c("--chrom_len_file"), action="store", dest="chrom_len_file", default=NULL, help="Chromosome lengths file"),
11 make_option(c("--ideas_input_config"), action="store", dest="ideas_input_config", help="Preprocessing output config file"), 12 make_option(c("--ideas_input_config"), action="store", dest="ideas_input_config", help="Preprocessing output config file"),
12 make_option(c("--ideaspre_input_config"), action="store", dest="ideaspre_input_config", help="Preprocessing input config file"), 13 make_option(c("--ideaspre_input_config"), action="store", dest="ideaspre_input_config", help="Preprocessing input config file"),
57 } else { 58 } else {
58 bigwig_file_name = file_path; 59 bigwig_file_name = file_path;
59 } 60 }
60 bed_file_name = paste(file_name, "bed", sep="."); 61 bed_file_name = paste(file_name, "bed", sep=".");
61 bed_file_path = paste("tmp", bed_file_name, sep="/"); 62 bed_file_path = paste("tmp", bed_file_name, sep="/");
62 cmd = paste("bigWigAverageOverBed", bigwig_file_name, opt$chrom_bed_input, "stdout | cut -f5 >", bed_file_path); 63 cmd = paste("bigWigAverageOverBed", bigwig_file_name, opt$bed_input, "stdout | cut -f5 >", bed_file_path);
63 system(cmd); 64 system(cmd);
64 cmd = paste("gzip -f", bed_file_path); 65 cmd = paste("gzip -f", bed_file_path);
65 system(cmd); 66 system(cmd);
66 } 67 }
67 68
68 # Create temporary file1.txt. 69 # Create file1.txt.
69 cmd = paste("cut -d' '", opt$ideaspre_input_config, "-f1,2 > file1.txt", sep=" "); 70 cmd = paste("cut -d' '", opt$ideaspre_input_config, "-f1,2 > file1.txt", sep=" ");
70 system(cmd); 71 system(cmd);
71 # Compress the bed files in the tmp directory. 72 # Compress the bed files in the tmp directory.
72 tmp_gzipped_files = paste(tmp_dir, "*.bed.gz", sep="/"); 73 tmp_gzipped_files = paste(tmp_dir, "*.bed.gz", sep="/");
73 # Create temporary file2.txt. 74 # Create file2.txt.
74 cmd = paste("ls", tmp_gzipped_files, "> file2.txt", sep=" "); 75 cmd = paste("ls", tmp_gzipped_files, "> file2.txt", sep=" ");
75 system(cmd); 76 system(cmd);
76 # Create IDEAS_input_config.txt with the format required by IDEAS. 77 # Create the ideas_input_config with the format required by IDEAS.
77 cmd = paste("paste -d' ' file1.txt file2.txt >", opt$ideas_input_config, sep=" "); 78 cmd = paste("paste -d' ' file1.txt file2.txt >", opt$ideas_input_config, sep=" ");
78 system(cmd); 79 system(cmd);
79 # Move IDEAS_input_config.txt to the output directory. 80 # Move the ideas_input_config to the output directory.
80 to_path = paste(opt$output_files_path, opt$ideas_input_config, sep="/"); 81 to_path = paste(opt$output_files_path, opt$ideas_input_config, sep="/");
81 file.rename(opt$ideas_input_config, to_path); 82 file.rename(opt$ideas_input_config, to_path);
82 # Move all compressed bed files in the tmp 83 # Move the compressed bed files in the tmp
83 # directory to the output tmp directory. 84 # directory to the output tmp directory.
84 tmp_files = list.files(path=tmp_dir); 85 tmp_files = list.files(path=tmp_dir);
85 for (i in 1:length(tmp_files)) { 86 for (i in 1:length(tmp_files)) {
86 from_path = paste(tmp_dir, tmp_files[i], sep="/"); 87 from_path = paste(tmp_dir, tmp_files[i], sep="/");
87 to_path = paste(output_tmp_dir, tmp_files[i], sep="/"); 88 to_path = paste(output_tmp_dir, tmp_files[i], sep="/");
88 file.rename(from_path, to_path); 89 file.rename(from_path, to_path);
89 } 90 }
90 if (!is.null(opt$chrom_bed_input) && !is.null(opt$chromosome_windows)) { 91 if (!is.null(opt$chromosome_windows)) {
91 # Rename chrom_bed_input to chromosomes.txt 92 # Move the chromosome_windows to the output directory.
92 # and move it to the output directory.
93 to_path = paste(opt$output_files_path, "chromosomes.bed", sep="/");
94 file.rename(opt$chrom_bed_input, to_path)
95 # Move chromosome_windows.txt to the output directory.
96 to_path = paste(opt$output_files_path, opt$chromosome_windows, sep="/"); 93 to_path = paste(opt$output_files_path, opt$chromosome_windows, sep="/");
97 file.rename(opt$chromosome_windows, to_path); 94 file.rename(opt$chromosome_windows, to_path);
98 } 95 }
99 # Create the primary HTML dataset. 96 # Create the primary HTML dataset.
100 create_primary_html(opt$output, opt$output_hid, opt$output_files_path); 97 create_primary_html(opt$output, opt$output_hid, opt$output_files_path);