annotate ideas_preprocessor.R @ 12:ab0f306504a3 draft

Uploaded
author greg
date Wed, 24 Jan 2018 13:47:23 -0500
parents f45e461b0557
children 20c21d946a8e
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1
91c5dbb14a13 Uploaded
greg
parents:
diff changeset
1 #!/usr/bin/env Rscript
91c5dbb14a13 Uploaded
greg
parents:
diff changeset
2
91c5dbb14a13 Uploaded
greg
parents:
diff changeset
3 suppressPackageStartupMessages(library("data.table"))
91c5dbb14a13 Uploaded
greg
parents:
diff changeset
4 suppressPackageStartupMessages(library("optparse"))
91c5dbb14a13 Uploaded
greg
parents:
diff changeset
5
91c5dbb14a13 Uploaded
greg
parents:
diff changeset
6 option_list <- list(
12
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
7 make_option(c("--chrom_bed_input"), action="store", dest="chrom_bed_input", defaul=NULL, help="Chromosome windows positions file"),
3
e97851e8951b Uploaded
greg
parents: 1
diff changeset
8 make_option(c("--exclude_input"), action="store", dest="exclude_input", defaul=NULL, help="File(s) containing regions to exclude"),
e97851e8951b Uploaded
greg
parents: 1
diff changeset
9 make_option(c("--bychr"), action="store_true", dest="bychr", defaul=FALSE, help="Separate files by chromosome"),
e97851e8951b Uploaded
greg
parents: 1
diff changeset
10 make_option(c("--chrom_len_file"), action="store", dest="chrom_len_file", default=NULL, help="Chromosome lengths file"),
5
2b4e1bd725f7 Uploaded
greg
parents: 3
diff changeset
11 make_option(c("--ideaspre_input_config"), action="store", dest="ideaspre_input_config", help="Preprocessing input config file"),
3
e97851e8951b Uploaded
greg
parents: 1
diff changeset
12 make_option(c("--output"), action="store", dest="output", help="Primary output dataset"),
e97851e8951b Uploaded
greg
parents: 1
diff changeset
13 make_option(c("--output_files_path"), action="store", dest="output_files_path", help="Primary output dataset extra files path"),
5
2b4e1bd725f7 Uploaded
greg
parents: 3
diff changeset
14 make_option(c("--output_hid"), action="store", dest="output_hid", help="Primary output dataset hid"),
3
e97851e8951b Uploaded
greg
parents: 1
diff changeset
15 make_option(c("--reads_per_bp"), action="store", dest="reads_per_bp", type="integer", help="Calculate the signal in each genomic window"),
e97851e8951b Uploaded
greg
parents: 1
diff changeset
16 make_option(c("--restrict_to_chroms"), action="store", dest="restrict_to_chroms", default=NULL, help="Restrict processing to specified chromosomes"),
e97851e8951b Uploaded
greg
parents: 1
diff changeset
17 make_option(c("--standardize_datasets"), action="store_true", dest="standardize_datasets", default=FALSE, help="Standardize all datasets"),
5
2b4e1bd725f7 Uploaded
greg
parents: 3
diff changeset
18 make_option(c("--chromosome_windows"), action="store", dest="chromosome_windows", default=NULL, help="Windows positions by chroms config file"),
3
e97851e8951b Uploaded
greg
parents: 1
diff changeset
19 make_option(c("--window_size"), action="store", dest="window_size", type="integer", default=NULL, help="Window size in base pairs")
1
91c5dbb14a13 Uploaded
greg
parents:
diff changeset
20 )
91c5dbb14a13 Uploaded
greg
parents:
diff changeset
21
91c5dbb14a13 Uploaded
greg
parents:
diff changeset
22 parser <- OptionParser(usage="%prog [options] file", option_list=option_list)
91c5dbb14a13 Uploaded
greg
parents:
diff changeset
23 args <- parse_args(parser, positional_arguments=TRUE)
91c5dbb14a13 Uploaded
greg
parents:
diff changeset
24 opt <- args$options
91c5dbb14a13 Uploaded
greg
parents:
diff changeset
25
5
2b4e1bd725f7 Uploaded
greg
parents: 3
diff changeset
26 create_primary_html = function(output, output_hid, output_files_path) {
2b4e1bd725f7 Uploaded
greg
parents: 3
diff changeset
27 files = list.files(path=output_files_path);
1
91c5dbb14a13 Uploaded
greg
parents:
diff changeset
28 s <- paste('<html><head></head><body>', sep="\n");
5
2b4e1bd725f7 Uploaded
greg
parents: 3
diff changeset
29 s <- paste(s, '<h3>History item ', output_hid, ' files prepared for IDEAS</h3>\n', sep="");
1
91c5dbb14a13 Uploaded
greg
parents:
diff changeset
30 s <- paste(s, '<ul>\n', sep="");
3
e97851e8951b Uploaded
greg
parents: 1
diff changeset
31 for (i in 1:length(files)) {
e97851e8951b Uploaded
greg
parents: 1
diff changeset
32 s <- paste(s, '<li><a href="', files[i], '">', files[i], '</a></li>\n', sep="");
1
91c5dbb14a13 Uploaded
greg
parents:
diff changeset
33 }
91c5dbb14a13 Uploaded
greg
parents:
diff changeset
34 s <- paste(s, '</ul>\n</body>\n</html>', sep="");
91c5dbb14a13 Uploaded
greg
parents:
diff changeset
35 cat(s, file=output);
91c5dbb14a13 Uploaded
greg
parents:
diff changeset
36 }
91c5dbb14a13 Uploaded
greg
parents:
diff changeset
37
3
e97851e8951b Uploaded
greg
parents: 1
diff changeset
38 tmp_dir = "tmp";
e97851e8951b Uploaded
greg
parents: 1
diff changeset
39 output_tmp_dir = paste(opt$output_files_path, tmp_dir, sep="/");
e97851e8951b Uploaded
greg
parents: 1
diff changeset
40 dir.create(output_tmp_dir, showWarnings=FALSE);
1
91c5dbb14a13 Uploaded
greg
parents:
diff changeset
41
5
2b4e1bd725f7 Uploaded
greg
parents: 3
diff changeset
42 # Read the ideaspre_input_config text file which has this format:
1
91c5dbb14a13 Uploaded
greg
parents:
diff changeset
43 # "cell type name" "epigenetic factor name" "file path" "file name" "datatype"
5
2b4e1bd725f7 Uploaded
greg
parents: 3
diff changeset
44 ideaspre_input_config = as.matrix(read.table(opt$ideaspre_input_config));
1
91c5dbb14a13 Uploaded
greg
parents:
diff changeset
45 # Process data to windows mean.
12
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
46 # TODO: implement scenario where user did not select chrom_bed_input.
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
47 if (!is.null(opt$chrom_bed_input)) {
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
48 for (i in 1:dim(ideaspre_input_config)[1]) {
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
49 file_path = ideaspre_input_config[i, 3]
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
50 file_name = ideaspre_input_config[i, 4]
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
51 datatype = ideaspre_input_config[i, 5]
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
52 if (datatype == "bam") {
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
53 cmd = paste("samtools index", file_path);
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
54 system(cmd);
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
55 bigwig_file_name = paste(file_name, "bw", sep=".");
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
56 cmd = paste("bamCoverage --bam", file_path, "-o", bigwig_file_name, "--binSize", opt$window_size);
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
57 system(cmd);
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
58 } else {
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
59 bigwig_file_name = file_path;
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
60 }
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
61 bed_file_name = paste(file_name, "bed", sep=".");
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
62 bed_file_path = paste("tmp", bed_file_name, sep="/");
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
63 cmd = paste("bigWigAverageOverBed", bigwig_file_name, opt$chrom_bed_input, "stdout | cut -f5 >", bed_file_path);
3
e97851e8951b Uploaded
greg
parents: 1
diff changeset
64 system(cmd);
12
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
65 cmd = paste("gzip -f", bed_file_path);
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
66 system(cmd);
1
91c5dbb14a13 Uploaded
greg
parents:
diff changeset
67 }
91c5dbb14a13 Uploaded
greg
parents:
diff changeset
68 }
91c5dbb14a13 Uploaded
greg
parents:
diff changeset
69
9
0cf23f33d13d Uploaded
greg
parents: 7
diff changeset
70 # Create file1.txt.
5
2b4e1bd725f7 Uploaded
greg
parents: 3
diff changeset
71 cmd = paste("cut -d' '", opt$ideaspre_input_config, "-f1,2 > file1.txt", sep=" ");
3
e97851e8951b Uploaded
greg
parents: 1
diff changeset
72 system(cmd);
e97851e8951b Uploaded
greg
parents: 1
diff changeset
73 # Compress the bed files in the tmp directory.
e97851e8951b Uploaded
greg
parents: 1
diff changeset
74 tmp_gzipped_files = paste(tmp_dir, "*.bed.gz", sep="/");
9
0cf23f33d13d Uploaded
greg
parents: 7
diff changeset
75 # Create file2.txt.
3
e97851e8951b Uploaded
greg
parents: 1
diff changeset
76 cmd = paste("ls", tmp_gzipped_files, "> file2.txt", sep=" ");
e97851e8951b Uploaded
greg
parents: 1
diff changeset
77 system(cmd);
12
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
78 # Create IDEAS_input_config.txt with the format required by IDEAS.
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
79 ideas_input_config = "IDEAS_input_config.txt"
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
80 cmd = paste("paste -d' ' file1.txt file2.txt >", ideas_input_config, sep=" " );
3
e97851e8951b Uploaded
greg
parents: 1
diff changeset
81 system(cmd);
12
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
82 # Move IDEAS_input_config.txt to the output directory.
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
83 to_path = paste(opt$output_files_path, ideas_input_config, sep="/");
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
84 file.rename(ideas_input_config, to_path);
9
0cf23f33d13d Uploaded
greg
parents: 7
diff changeset
85 # Move the compressed bed files in the tmp
3
e97851e8951b Uploaded
greg
parents: 1
diff changeset
86 # directory to the output tmp directory.
e97851e8951b Uploaded
greg
parents: 1
diff changeset
87 tmp_files = list.files(path=tmp_dir);
e97851e8951b Uploaded
greg
parents: 1
diff changeset
88 for (i in 1:length(tmp_files)) {
e97851e8951b Uploaded
greg
parents: 1
diff changeset
89 from_path = paste(tmp_dir, tmp_files[i], sep="/");
e97851e8951b Uploaded
greg
parents: 1
diff changeset
90 to_path = paste(output_tmp_dir, tmp_files[i], sep="/");
e97851e8951b Uploaded
greg
parents: 1
diff changeset
91 file.rename(from_path, to_path);
e97851e8951b Uploaded
greg
parents: 1
diff changeset
92 }
12
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
93 if (!is.null(opt$chrom_bed_input) && !is.null(opt$chromosome_windows)) {
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
94 # Renane opt$chrom_bed_input to be chromosomes.bed
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
95 # and make a copy of it in the output directory.
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
96 to_path = paste(opt$output_files_path, "chromosomes.bed", sep="/");
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
97 file.copy(opt$chromosome_windows, to_path);
ab0f306504a3 Uploaded
greg
parents: 10
diff changeset
98 # Move chromosome_windows.txt to the output directory.
5
2b4e1bd725f7 Uploaded
greg
parents: 3
diff changeset
99 to_path = paste(opt$output_files_path, opt$chromosome_windows, sep="/");
2b4e1bd725f7 Uploaded
greg
parents: 3
diff changeset
100 file.rename(opt$chromosome_windows, to_path);
3
e97851e8951b Uploaded
greg
parents: 1
diff changeset
101 }
1
91c5dbb14a13 Uploaded
greg
parents:
diff changeset
102 # Create the primary HTML dataset.
9
0cf23f33d13d Uploaded
greg
parents: 7
diff changeset
103 create_primary_html(opt$output, opt$output_hid, opt$output_files_path);