annotate ideas_genome_tracks.R @ 60:8baf6aa848d1 draft

Uploaded
author greg
date Tue, 19 Dec 2017 09:11:35 -0500
parents 1736d09a6927
children 5acfb82e4f44
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
1 #!/usr/bin/env Rscript
a842ab7e3bba Uploaded
greg
parents:
diff changeset
2
5
d1703a15c393 Uploaded
greg
parents: 0
diff changeset
3 suppressPackageStartupMessages(library("data.table"))
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
4 suppressPackageStartupMessages(library("optparse"))
47
c31d53123fae Uploaded
greg
parents: 45
diff changeset
5 suppressPackageStartupMessages(library("viridisLite"))
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
6
a842ab7e3bba Uploaded
greg
parents:
diff changeset
7 option_list <- list(
54
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
8 make_option(c("--build"), action="store", dest="build", help="Genome build"),
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
9 make_option(c("--chrom_len_file"), action="store", dest="chrom_len_file", help="Chromosome length file"),
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
10 make_option(c("--email"), action="store", dest="email", help="User email address"),
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
11 make_option(c("--galaxy_url"), action="store", dest="galaxy_url", help="Galaxy instance base URL"),
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
12 make_option(c("--hub_name"), action="store", dest="hub_name", default=NULL, help="Hub name without spaces"),
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
13 make_option(c("--input_dir_state"), action="store", dest="input_dir_state", help="Directory containing .state outputs from IDEAS"),
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
14 make_option(c("--long_label"), action="store", dest="long_label", help="Hub long label"),
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
15 make_option(c("--output_trackhub"), action="store", dest="output_trackhub", help="Output hub file"),
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
16 make_option(c("--output_trackhub_files_path"), action="store", dest="output_trackhub_files_path", help="Output hub extra files path"),
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
17 make_option(c("--output_trackhub_id"), action="store", dest="output_trackhub_id", help="Encoded output_trackhub dataset id"),
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
18 make_option(c("--short_label"), action="store", dest="short_label", help="Hub short label")
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
19 )
a842ab7e3bba Uploaded
greg
parents:
diff changeset
20
a842ab7e3bba Uploaded
greg
parents:
diff changeset
21 parser <- OptionParser(usage="%prog [options] file", option_list=option_list)
a842ab7e3bba Uploaded
greg
parents:
diff changeset
22 args <- parse_args(parser, positional_arguments=TRUE)
a842ab7e3bba Uploaded
greg
parents:
diff changeset
23 opt <- args$options
a842ab7e3bba Uploaded
greg
parents:
diff changeset
24
39
ecfe907615c5 Uploaded
greg
parents: 37
diff changeset
25 create_primary_html = function(output_trackhub, tracks_dir, build) {
54
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
26 track_files <- list.files(path=tracks_dir);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
27 s <- paste('<html><head></head><body>', sep="\n");
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
28 s <- paste(s, '<h3>Contents of directory ~/myHub/', build, ' required by UCSC TrackHub</h3>\n', sep="");
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
29 s <- paste(s, '<ul>\n', sep="")
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
30 for (i in 1:length(track_files)) {
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
31 s <- paste(s, '<li><a href="', 'myHub/', build, "/", track_files[i], '">', track_files[i], '</a></li>\n', sep="");
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
32 }
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
33 s <- paste(s, '</ul>\n</body>\n</html>', sep="");
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
34 cat(s, file=output_trackhub);
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
35 }
696e435f5607 Uploaded
greg
parents: 18
diff changeset
36
51
f1045dd36961 Uploaded
greg
parents: 50
diff changeset
37 create_track = function(input_dir_state, chrom_len_file, base_track_file_name) {
54
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
38 # Create everythin needed, including the bigbed file,
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
39 # to render the tracks within the UCSC track hub.
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
40 state_files <- list.files(path=input_dir_state, full.names=TRUE);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
41 genome_size = read.table(chrom_len_file);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
42 g = NULL;
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
43 for(i in state_files) {
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
44 tg = as.matrix(fread(i));
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
45 t = NULL;
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
46 for(j in 1:dim(genome_size)[1]) {
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
47 t = c(t, which(tg[,2]==as.character(genome_size[j, 1]) & as.numeric(tg[,4]) > as.numeric(genome_size[j, 2])));
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
48 }
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
49 if (length(t) > 0) {
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
50 tg = tg[-t,];
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
51 }
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
52 t = which(is.na(match(tg[,2], genome_size[,1]))==T);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
53 if (length(t)>0) {
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
54 tg = tg[-t,];
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
55 }
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
56 g = rbind(g, tg);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
57 }
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
58 uchr = sort(unique(as.character(g[,2])));
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
59 g1 = NULL;
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
60 for(i in uchr) {
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
61 t = which(g[,2]==i);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
62 g1 = rbind(g1, g[t[order(as.integer(g[t, 3]))],]);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
63 }
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
64 g = NULL;
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
65 chr = as.character(g1[,2]);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
66 posst = as.numeric(g1[,3]);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
67 posed = as.numeric(g1[,4]);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
68 state = as.matrix(g1[,5:(dim(g1)[2]-1)]);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
69 state_name = 0:max(state);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
70 L = dim(g1)[1];
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
71 n = dim(state)[2];
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
72 cells = as.character(colnames(g1)[5:(dim(g1)[2]-1)]);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
73 options(scipen=999);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
74 tt = which(chr[2:L]!=chr[2:L-1]);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
75 tt = c(tt, which(posst[2:L]!=posed[2:L-1]));
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
76 tt = sort(unique(tt));
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
77 for(i in 1:n) {
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
78 tstate = state[,i];
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
79 t = c(tt, which(tstate[2:L]!=tstate[2:L-1]));
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
80 t = sort(unique(t));
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
81 t0 = c(0, t) + 1;
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
82 t = c(t, L);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
83 np = cbind(chr[t], posst[t0], posed[t], tstate[t]);
57
7a7d5c3c5480 Uploaded
greg
parents: 56
diff changeset
84 track_file_name_bed_unsorted <- get_track_file_name(base_track_file_name, i, "bed_unsorted");
58
269241d4b9c8 Uploaded
greg
parents: 57
diff changeset
85 track_file_name_bed <- get_track_file_name(base_track_file_name, i, "bed");
56
c54ef43eaee3 Uploaded
greg
parents: 55
diff changeset
86 track_file_name_bigbed <- get_track_file_name(base_track_file_name, i, "bigbed");
54
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
87 x = cbind(np[, 1:3], state_name[as.integer(np[,4])+1], 1000, ".", np[,2:3]);
57
7a7d5c3c5480 Uploaded
greg
parents: 56
diff changeset
88 write.table(as.matrix(x), track_file_name_bed_unsorted, quote=F, row.names=F, col.names=F);
59
1736d09a6927 Uploaded
greg
parents: 58
diff changeset
89 cmd = paste("LC_COLLATE=C sort -k1,1 -k2,2n < ", track_file_name_bed_unsorted, " > ", track_file_name_bed);
1736d09a6927 Uploaded
greg
parents: 58
diff changeset
90 system(cmd);
1736d09a6927 Uploaded
greg
parents: 58
diff changeset
91 cmd = paste("bedToBigBed ", track_file_name_bed, chrom_len_file, " ", track_file_name_bigbed);
1736d09a6927 Uploaded
greg
parents: 58
diff changeset
92 system(cmd);
58
269241d4b9c8 Uploaded
greg
parents: 57
diff changeset
93 system(paste("rm ", track_file_name_bed_unsorted));
269241d4b9c8 Uploaded
greg
parents: 57
diff changeset
94 system(paste("rm ", track_file_name_bed));
54
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
95 }
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
96 return(cells);
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
97 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
98
51
f1045dd36961 Uploaded
greg
parents: 50
diff changeset
99 create_track_db = function(galaxy_url, encoded_dataset_id, input_dir_state, build, chrom_len_file, tracks_dir, hub_name, short_label, long_label) {
54
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
100 # Create a trackDb.txt file that includes each state.
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
101 base_track_file_name <- paste(tracks_dir, hub_name, sep="");
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
102 cells = create_track(input_dir_state, chrom_len_file, base_track_file_name);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
103 # Create a a character vector of 1024 viridis color hex codes.
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
104 viridis_vector <- viridis(1024, alpha=1, begin=0, end=1, direction=1, option="D");
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
105 colors_used <- vector();
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
106 track_db = NULL;
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
107 for (i in 1:length(cells)) {
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
108 # Generate a random number between 1 and 1024 as
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
109 # the viridis_vector index for the next state color.
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
110 color_index <- sample(1:1024, 1)
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
111 # Make sure the color was not previously chosen.
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
112 while(is.element(color_index, colors_used)) {
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
113 color_index <- sample(1:1024, 1)
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
114 }
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
115 # Append the color to our list of chosen colors.
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
116 append(colors_used, color_index)
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
117 # Get the hex code from viridis_vector.
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
118 viridis_color_hex_code <- viridis_vector[color_index]
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
119 big_data_url <- get_big_data_url(galaxy_url, encoded_dataset_id, tracks_dir, i, build);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
120 # trackDb.txt track entry.
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
121 track_db = c(track_db, paste("track ", hub_name, "_track_", i, sep=""));
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
122 track_db = c(track_db, "type bigBed");
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
123 track_db = c(track_db, paste("bigDataUrl", big_data_url, sep=" "));
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
124 track_db = c(track_db, paste("shortLabel", short_label, sep=" "));
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
125 track_db = c(track_db, paste("longLabel", long_label, sep=" "));
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
126 track_db = c(track_db, paste("priority", i));
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
127 track_db = c(track_db, "itemRgb on");
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
128 track_db = c(track_db, "maxItems 100000");
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
129 track_db = c(track_db, paste("color", paste(c(col2rgb(viridis_color_hex_code)), collapse=","), sep=" "));
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
130 track_db = c(track_db, "visibility dense");
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
131 track_db = c(track_db, "");
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
132 }
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
133 return(track_db);
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
134 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
135
39
ecfe907615c5 Uploaded
greg
parents: 37
diff changeset
136 get_big_data_url = function(galaxy_url, encoded_dataset_id, tracks_dir, index, build) {
56
c54ef43eaee3 Uploaded
greg
parents: 55
diff changeset
137 track_files <- list.files(path=tracks_dir, pattern="\\.bigbed");
54
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
138 s <- paste(galaxy_url, 'datasets/', encoded_dataset_id, '/display/myHub/', build, '/', track_files[index], sep="");
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
139 return(s)
50
0c66347d5354 Uploaded
greg
parents: 48
diff changeset
140 }
0c66347d5354 Uploaded
greg
parents: 48
diff changeset
141
39
ecfe907615c5 Uploaded
greg
parents: 37
diff changeset
142 get_track_file_name = function(base_track_file_name, index, ext) {
54
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
143 track_file_name <- paste(base_track_file_name, index, ext, sep=".");
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
144 return(track_file_name);
39
ecfe907615c5 Uploaded
greg
parents: 37
diff changeset
145 }
ecfe907615c5 Uploaded
greg
parents: 37
diff changeset
146
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
147 # Create the hub.txt output.
53
8b1c49f2ecaa Uploaded
greg
parents: 52
diff changeset
148 hub_name_line <- paste("hub ", opt$hub_name, sep="");
39
ecfe907615c5 Uploaded
greg
parents: 37
diff changeset
149 short_label_line <- paste("shortLabel ", opt$short_label, sep="");
ecfe907615c5 Uploaded
greg
parents: 37
diff changeset
150 long_label_line <- paste("longLabel ", opt$long_label, sep="");
35
3e001c30b0f7 Uploaded
greg
parents: 34
diff changeset
151 genomes_txt_line <- paste("genomesFile genomes.txt", sep="");
39
ecfe907615c5 Uploaded
greg
parents: 37
diff changeset
152 email_line <- paste("email ", opt$email, sep="");
ecfe907615c5 Uploaded
greg
parents: 37
diff changeset
153 contents <- paste(hub_name_line, short_label_line, long_label_line, genomes_txt_line, email_line, sep="\n");
35
3e001c30b0f7 Uploaded
greg
parents: 34
diff changeset
154 hub_dir <- paste(opt$output_trackhub_files_path, "/", "myHub", "/", sep="");
22
a3740127fc5e Uploaded
greg
parents: 20
diff changeset
155 dir.create(hub_dir, showWarnings=FALSE);
35
3e001c30b0f7 Uploaded
greg
parents: 34
diff changeset
156 hub_file_path <- paste(hub_dir, "hub.txt", sep="");
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
157 write.table(contents, file=hub_file_path, quote=F, row.names=F, col.names=F);
696e435f5607 Uploaded
greg
parents: 18
diff changeset
158
696e435f5607 Uploaded
greg
parents: 18
diff changeset
159 # Create the genomes.txt output.
39
ecfe907615c5 Uploaded
greg
parents: 37
diff changeset
160 genome_line <- paste("genome ", opt$build, sep="");
35
3e001c30b0f7 Uploaded
greg
parents: 34
diff changeset
161 track_db_line <- paste("trackDb ", opt$build, "/", "trackDb.txt", sep="");
39
ecfe907615c5 Uploaded
greg
parents: 37
diff changeset
162 contents <- paste(genome_line, track_db_line, sep="\n");
30
4c42188478a7 Uploaded
greg
parents: 29
diff changeset
163 genomes_file_path <- paste(hub_dir, "genomes.txt", sep="");
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
164 write.table(contents, file=genomes_file_path, quote=F, row.names=F, col.names=F);
696e435f5607 Uploaded
greg
parents: 18
diff changeset
165
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
166 # Create the tracks.
36
fe9dfd95f13d Uploaded
greg
parents: 35
diff changeset
167 tracks_dir <- paste(hub_dir, opt$build, "/", sep="");
11
45a2f84b8247 Uploaded
greg
parents: 9
diff changeset
168 dir.create(tracks_dir, showWarnings=FALSE);
51
f1045dd36961 Uploaded
greg
parents: 50
diff changeset
169 track_db <- create_track_db(opt$galaxy_url, opt$output_trackhub_id, opt$input_dir_state, opt$build, opt$chrom_len_file, tracks_dir, opt$hub_name, opt$short_label, opt$long_label);
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
170
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
171 # Create the trackDb.txt output.
35
3e001c30b0f7 Uploaded
greg
parents: 34
diff changeset
172 track_db_file_path <- paste(tracks_dir, "trackDb.txt", sep="");
13
f0dfff1128c8 Uploaded
greg
parents: 12
diff changeset
173 write.table(track_db, file=track_db_file_path, quote=F, row.names=F, col.names=F);
11
45a2f84b8247 Uploaded
greg
parents: 9
diff changeset
174
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
175 # Create the primary HTML dataset.
54
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
176 create_primary_html(opt$output_trackhub, tracks_dir, opt$build);