|
5
|
1 (.venv) [galaxy@IDEAS ideas_preprocessor]$ cat ideas_preprocessor.R
|
|
1
|
2 #!/usr/bin/env Rscript
|
|
|
3
|
|
|
4 suppressPackageStartupMessages(library("data.table"))
|
|
|
5 suppressPackageStartupMessages(library("optparse"))
|
|
|
6
|
|
|
7 option_list <- list(
|
|
5
|
8 make_option(c("--chrom_bed_input"), action="store", dest="chrom_bed_input", defaul=NULL, help="Chromosome windows positions file"),
|
|
3
|
9 make_option(c("--exclude_input"), action="store", dest="exclude_input", defaul=NULL, help="File(s) containing regions to exclude"),
|
|
|
10 make_option(c("--bychr"), action="store_true", dest="bychr", defaul=FALSE, help="Separate files by chromosome"),
|
|
|
11 make_option(c("--chrom_len_file"), action="store", dest="chrom_len_file", default=NULL, help="Chromosome lengths file"),
|
|
5
|
12 make_option(c("--ideas_input_config"), action="store", dest="ideas_input_config", help="Preprocessing output config file"),
|
|
|
13 make_option(c("--ideaspre_input_config"), action="store", dest="ideaspre_input_config", help="Preprocessing input config file"),
|
|
3
|
14 make_option(c("--output"), action="store", dest="output", help="Primary output dataset"),
|
|
|
15 make_option(c("--output_files_path"), action="store", dest="output_files_path", help="Primary output dataset extra files path"),
|
|
5
|
16 make_option(c("--output_hid"), action="store", dest="output_hid", help="Primary output dataset hid"),
|
|
3
|
17 make_option(c("--reads_per_bp"), action="store", dest="reads_per_bp", type="integer", help="Calculate the signal in each genomic window"),
|
|
|
18 make_option(c("--restrict_to_chroms"), action="store", dest="restrict_to_chroms", default=NULL, help="Restrict processing to specified chromosomes"),
|
|
|
19 make_option(c("--standardize_datasets"), action="store_true", dest="standardize_datasets", default=FALSE, help="Standardize all datasets"),
|
|
5
|
20 make_option(c("--chromosome_windows"), action="store", dest="chromosome_windows", default=NULL, help="Windows positions by chroms config file"),
|
|
3
|
21 make_option(c("--window_size"), action="store", dest="window_size", type="integer", default=NULL, help="Window size in base pairs")
|
|
1
|
22 )
|
|
|
23
|
|
|
24 parser <- OptionParser(usage="%prog [options] file", option_list=option_list)
|
|
|
25 args <- parse_args(parser, positional_arguments=TRUE)
|
|
|
26 opt <- args$options
|
|
|
27
|
|
5
|
28 create_primary_html = function(output, output_hid, output_files_path) {
|
|
|
29 files = list.files(path=output_files_path);
|
|
1
|
30 s <- paste('<html><head></head><body>', sep="\n");
|
|
5
|
31 s <- paste(s, '<h3>History item ', output_hid, ' files prepared for IDEAS</h3>\n', sep="");
|
|
1
|
32 s <- paste(s, '<ul>\n', sep="");
|
|
3
|
33 for (i in 1:length(files)) {
|
|
|
34 s <- paste(s, '<li><a href="', files[i], '">', files[i], '</a></li>\n', sep="");
|
|
1
|
35 }
|
|
|
36 s <- paste(s, '</ul>\n</body>\n</html>', sep="");
|
|
|
37 cat(s, file=output);
|
|
|
38 }
|
|
|
39
|
|
3
|
40 tmp_dir = "tmp";
|
|
|
41 output_tmp_dir = paste(opt$output_files_path, tmp_dir, sep="/");
|
|
|
42 dir.create(output_tmp_dir, showWarnings=FALSE);
|
|
1
|
43
|
|
5
|
44 # Read the ideaspre_input_config text file which has this format:
|
|
1
|
45 # "cell type name" "epigenetic factor name" "file path" "file name" "datatype"
|
|
5
|
46 ideaspre_input_config = as.matrix(read.table(opt$ideaspre_input_config));
|
|
1
|
47 # Process data to windows mean.
|
|
5
|
48 for (i in 1:dim(ideaspre_input_config)[1]) {
|
|
|
49 file_path = ideaspre_input_config[i, 3]
|
|
|
50 file_name = ideaspre_input_config[i, 4]
|
|
|
51 datatype = ideaspre_input_config[i, 5]
|
|
1
|
52 if (datatype == "bam") {
|
|
3
|
53 cmd = paste("samtools index", file_path);
|
|
|
54 system(cmd);
|
|
|
55 bigwig_file_name = paste(file_name, "bw", sep=".");
|
|
|
56 cmd = paste("bamCoverage --bam", file_path, "-o", bigwig_file_name, "--binSize", opt$window_size);
|
|
|
57 system(cmd);
|
|
1
|
58 } else {
|
|
3
|
59 bigwig_file_name = file_path;
|
|
1
|
60 }
|
|
3
|
61 bed_file_name = paste(file_name, "bed", sep=".");
|
|
|
62 bed_file_path = paste("tmp", bed_file_name, sep="/");
|
|
5
|
63 cmd = paste("bigWigAverageOverBed", bigwig_file_name, opt$chrom_bed_input, "stdout | cut -f5 >", bed_file_path);
|
|
3
|
64 system(cmd);
|
|
|
65 cmd = paste("gzip -f", bed_file_path);
|
|
|
66 system(cmd);
|
|
1
|
67 }
|
|
|
68
|
|
5
|
69 # Create temporary file1.txt.
|
|
|
70 cmd = paste("cut -d' '", opt$ideaspre_input_config, "-f1,2 > file1.txt", sep=" ");
|
|
3
|
71 system(cmd);
|
|
|
72 # Compress the bed files in the tmp directory.
|
|
|
73 tmp_gzipped_files = paste(tmp_dir, "*.bed.gz", sep="/");
|
|
5
|
74 # Create temporary file2.txt.
|
|
3
|
75 cmd = paste("ls", tmp_gzipped_files, "> file2.txt", sep=" ");
|
|
|
76 system(cmd);
|
|
5
|
77 # Create IDEAS_input_config.txt with the format required by IDEAS.
|
|
|
78 cmd = paste("paste -d' ' file1.txt file2.txt >", opt$ideas_input_config, sep=" ");
|
|
3
|
79 system(cmd);
|
|
5
|
80 # Move IDEAS_input_config.txt to the output directory.
|
|
|
81 to_path = paste(opt$output_files_path, opt$ideas_input_config, sep="/");
|
|
|
82 file.rename(opt$ideas_input_config, to_path);
|
|
|
83 # Move all compressed bed files in the tmp
|
|
3
|
84 # directory to the output tmp directory.
|
|
|
85 tmp_files = list.files(path=tmp_dir);
|
|
|
86 for (i in 1:length(tmp_files)) {
|
|
|
87 from_path = paste(tmp_dir, tmp_files[i], sep="/");
|
|
|
88 to_path = paste(output_tmp_dir, tmp_files[i], sep="/");
|
|
|
89 file.rename(from_path, to_path);
|
|
|
90 }
|
|
5
|
91 if (!is.nulll(opt$chrom_bed_input) && !is.null(opt$chromosome_windows)) {
|
|
|
92 # Rename chrom_bed_input to chromosomes.txt
|
|
|
93 # and move it to the output directory.
|
|
|
94 to_path = paste(opt$output_files_path, "chromosomes.bed", sep="/");
|
|
|
95 file.rename(opt$chrom_bed_input, to_path)
|
|
|
96 # Move chromosome_windows.txt to the output directory.
|
|
|
97 to_path = paste(opt$output_files_path, opt$chromosome_windows, sep="/");
|
|
|
98 file.rename(opt$chromosome_windows, to_path);
|
|
3
|
99 }
|
|
1
|
100 # Create the primary HTML dataset.
|
|
5
|
101 create_primary_html(opt$output, opt$output_hid, opt$output_files_path);
|
