annotate ideas_genome_tracks.R @ 83:2f49b6bddccc draft

Uploaded
author greg
date Thu, 04 Jan 2018 14:44:38 -0500
parents 6d97ef6bf7b7
children 9b43f5773db9
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"))
a842ab7e3bba Uploaded
greg
parents:
diff changeset
5
a842ab7e3bba Uploaded
greg
parents:
diff changeset
6 option_list <- list(
83
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
7 make_option(c("--build"), action="store", dest="build", help="Genome build"),
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
8 make_option(c("--chrom_len_file"), action="store", dest="chrom_len_file", help="Chromosome length file"),
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
9 make_option(c("--email"), action="store", dest="email", help="User email address"),
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
10 make_option(c("--galaxy_url"), action="store", dest="galaxy_url", help="Galaxy instance base URL"),
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
11 make_option(c("--hub_long_label"), action="store", dest="hub_long_label", help="Hub long label"),
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
12 make_option(c("--hub_name"), action="store", dest="hub_name", default=NULL, help="Hub name without spaces"),
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
13 make_option(c("--hub_short_label"), action="store", dest="hub_short_label", help="Hub short label"),
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
14 make_option(c("--input_dir_para"), action="store", dest="input_dir_para", help="Directory containing .para outputs from IDEAS"),
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
15 make_option(c("--input_dir_state"), action="store", dest="input_dir_state", help="Directory containing .state outputs from IDEAS"),
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
16 make_option(c("--output_trackhub"), action="store", dest="output_trackhub", help="Output hub file"),
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
17 make_option(c("--output_trackhub_files_path"), action="store", dest="output_trackhub_files_path", help="Output hub extra files path"),
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
18 make_option(c("--output_trackhub_id"), action="store", dest="output_trackhub_id", help="Encoded output_trackhub dataset id"),
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
19 make_option(c("--script_dir"), action="store", dest="script_dir", help="R script source directory"),
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
20 make_option(c("--state_colors"), action="store", dest="state_colors", default=NULL, help="List of state colors"),
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
21 make_option(c("--state_indexes"), action="store", dest="state_indexes", default=NULL, help="List of state indexes"),
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
22 make_option(c("--state_names"), action="store", dest="state_names", default=NULL, help="List of state names")
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
23 )
a842ab7e3bba Uploaded
greg
parents:
diff changeset
24
a842ab7e3bba Uploaded
greg
parents:
diff changeset
25 parser <- OptionParser(usage="%prog [options] file", option_list=option_list)
a842ab7e3bba Uploaded
greg
parents:
diff changeset
26 args <- parse_args(parser, positional_arguments=TRUE)
a842ab7e3bba Uploaded
greg
parents:
diff changeset
27 opt <- args$options
a842ab7e3bba Uploaded
greg
parents:
diff changeset
28
78
55da56869a0d Uploaded
greg
parents: 77
diff changeset
29 create_primary_html = function(output_trackhub, tracks_dir, build) {
55da56869a0d Uploaded
greg
parents: 77
diff changeset
30 track_files <- list.files(path=tracks_dir);
54
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
31 s <- paste('<html><head></head><body>', sep="\n");
78
55da56869a0d Uploaded
greg
parents: 77
diff changeset
32 s <- paste(s, '<h3>Contents of directory ~/myHub/', build, ' required by UCSC TrackHub</h3>\n', sep="");
54
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
33 s <- paste(s, '<ul>\n', sep="")
78
55da56869a0d Uploaded
greg
parents: 77
diff changeset
34 for (i in 1:length(track_files)) {
55da56869a0d Uploaded
greg
parents: 77
diff changeset
35 s <- paste(s, '<li><a href="', 'myHub/', build, "/", track_files[i], '">', track_files[i], '</a></li>\n', sep="");
54
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
36 }
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
37 s <- paste(s, '</ul>\n</body>\n</html>', sep="");
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
38 cat(s, file=output_trackhub);
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
39 }
696e435f5607 Uploaded
greg
parents: 18
diff changeset
40
77
8c4549f123da Uploaded
greg
parents: 75
diff changeset
41 create_cells = function(input_dir_state, chrom_len_file, base_track_file_name, state_indexes, state_names, state_colors) {
66
f0c488b73dbc Uploaded
greg
parents: 65
diff changeset
42 # Create everything needed, including the bigbed file,
54
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
43 # to render the tracks within the UCSC track hub.
83
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
44 state_files = list.files(path=input_dir_state, full.names=TRUE);
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
45 cell_type_names = get_cell_type_names(state_files[1]);
54
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
46 genome_size = read.table(chrom_len_file);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
47 g = NULL;
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
48 for(i in state_files) {
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
49 tg = as.matrix(fread(i));
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
50 t = NULL;
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
51 for(j in 1:dim(genome_size)[1]) {
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
52 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
53 }
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
54 if (length(t) > 0) {
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
55 tg = tg[-t,];
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
56 }
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
57 t = which(is.na(match(tg[,2], genome_size[,1]))==T);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
58 if (length(t)>0) {
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
59 tg = tg[-t,];
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
60 }
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
61 g = rbind(g, tg);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
62 }
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
63 uchr = sort(unique(as.character(g[,2])));
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
64 g1 = NULL;
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
65 for(i in uchr) {
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
66 t = which(g[,2]==i);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
67 g1 = rbind(g1, g[t[order(as.integer(g[t, 3]))],]);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
68 }
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
69 g = NULL;
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
70 chr = as.character(g1[,2]);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
71 posst = as.numeric(g1[,3]);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
72 posed = as.numeric(g1[,4]);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
73 state = as.matrix(g1[,5:(dim(g1)[2]-1)]);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
74 L = dim(g1)[1];
83
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
75 # Here n will be the same length as the
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
76 # list of cell_type_names defined above.
54
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
77 n = dim(state)[2];
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
78 cells = as.character(colnames(g1)[5:(dim(g1)[2]-1)]);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
79 options(scipen=999);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
80 tt = which(chr[2:L]!=chr[2:L-1]);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
81 tt = c(tt, which(posst[2:L]!=posed[2:L-1]));
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
82 tt = sort(unique(tt));
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
83 for(i in 1:n) {
83
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
84 state_name = cell_type_names[i];
54
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
85 tstate = state[,i];
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
86 t = c(tt, which(tstate[2:L]!=tstate[2:L-1]));
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
87 t = sort(unique(t));
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
88 t0 = c(0, t) + 1;
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
89 t = c(t, L);
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
90 np = cbind(chr[t], posst[t0], posed[t], tstate[t]);
57
7a7d5c3c5480 Uploaded
greg
parents: 56
diff changeset
91 track_file_name_bed_unsorted <- get_track_file_name(base_track_file_name, i, "bed_unsorted");
58
269241d4b9c8 Uploaded
greg
parents: 57
diff changeset
92 track_file_name_bed <- get_track_file_name(base_track_file_name, i, "bed");
56
c54ef43eaee3 Uploaded
greg
parents: 55
diff changeset
93 track_file_name_bigbed <- get_track_file_name(base_track_file_name, i, "bigbed");
83
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
94 x = cbind(np[, 1:3], state_name, 1000, ".", np[,2:3]);
57
7a7d5c3c5480 Uploaded
greg
parents: 56
diff changeset
95 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
96 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
97 system(cmd);
1736d09a6927 Uploaded
greg
parents: 58
diff changeset
98 cmd = paste("bedToBigBed ", track_file_name_bed, chrom_len_file, " ", track_file_name_bigbed);
1736d09a6927 Uploaded
greg
parents: 58
diff changeset
99 system(cmd);
65
794d2104f83d Uploaded
greg
parents: 64
diff changeset
100 system(paste("rm ", track_file_name_bed_unsorted));
794d2104f83d Uploaded
greg
parents: 64
diff changeset
101 system(paste("rm ", track_file_name_bed));
54
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
102 }
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
103 return(cells);
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
104 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
105
77
8c4549f123da Uploaded
greg
parents: 75
diff changeset
106 create_track_db = function(galaxy_url, encoded_dataset_id, input_dir_para, input_dir_state, build, chrom_len_file,
8c4549f123da Uploaded
greg
parents: 75
diff changeset
107 tracks_dir, hub_name, hub_short_label, hub_long_label, state_indexes, state_names, state_colors) {
54
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
108 # Create a trackDb.txt file that includes each state.
80
6d97ef6bf7b7 Uploaded
greg
parents: 78
diff changeset
109 s_indexes = c();
6d97ef6bf7b7 Uploaded
greg
parents: 78
diff changeset
110 s_names = c();
6d97ef6bf7b7 Uploaded
greg
parents: 78
diff changeset
111 s_colors = c();
6d97ef6bf7b7 Uploaded
greg
parents: 78
diff changeset
112 if (is.null(state_indexes)) {
6d97ef6bf7b7 Uploaded
greg
parents: 78
diff changeset
113 # Generate state colors automatically.
6d97ef6bf7b7 Uploaded
greg
parents: 78
diff changeset
114 index = 0;
6d97ef6bf7b7 Uploaded
greg
parents: 78
diff changeset
115 para_files = list.files(path=input_dir_para, full.names=TRUE);
6d97ef6bf7b7 Uploaded
greg
parents: 78
diff changeset
116 for (para_file in para_files) {
6d97ef6bf7b7 Uploaded
greg
parents: 78
diff changeset
117 data_frame = read.table(para_file, comment="!", header=T);
6d97ef6bf7b7 Uploaded
greg
parents: 78
diff changeset
118 color_hex_code = create_heatmap(data_frame);
6d97ef6bf7b7 Uploaded
greg
parents: 78
diff changeset
119 s_indexes = c(s_indexes, index);
6d97ef6bf7b7 Uploaded
greg
parents: 78
diff changeset
120 s_names = c(s_names, toString(index));
6d97ef6bf7b7 Uploaded
greg
parents: 78
diff changeset
121 s_colors = c(s_colors, color_hex_code);
6d97ef6bf7b7 Uploaded
greg
parents: 78
diff changeset
122 }
6d97ef6bf7b7 Uploaded
greg
parents: 78
diff changeset
123 } else {
61
5acfb82e4f44 Uploaded
greg
parents: 59
diff changeset
124 # Split state_indexes into a list of integers.
80
6d97ef6bf7b7 Uploaded
greg
parents: 78
diff changeset
125 index_str = as.character(state_indexes);
6d97ef6bf7b7 Uploaded
greg
parents: 78
diff changeset
126 items = strsplit(index_str, ",")
61
5acfb82e4f44 Uploaded
greg
parents: 59
diff changeset
127 for (item in items) {
80
6d97ef6bf7b7 Uploaded
greg
parents: 78
diff changeset
128 s_indexes = c(s_indexes, as.integer(item));
61
5acfb82e4f44 Uploaded
greg
parents: 59
diff changeset
129 }
77
8c4549f123da Uploaded
greg
parents: 75
diff changeset
130 # Split state_names into a list of strings.
83
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
131 name_str = as.character(state_names);
80
6d97ef6bf7b7 Uploaded
greg
parents: 78
diff changeset
132 items = strsplit(name_str, ",");
6d97ef6bf7b7 Uploaded
greg
parents: 78
diff changeset
133 item_index = 0;
77
8c4549f123da Uploaded
greg
parents: 75
diff changeset
134 for (item in items) {
8c4549f123da Uploaded
greg
parents: 75
diff changeset
135 item_index = item_index + 1;
8c4549f123da Uploaded
greg
parents: 75
diff changeset
136 # Handle the special string "use state index".
8c4549f123da Uploaded
greg
parents: 75
diff changeset
137 if (identical(item, "use state index")) {
80
6d97ef6bf7b7 Uploaded
greg
parents: 78
diff changeset
138 item = s_indexes[item_index];
77
8c4549f123da Uploaded
greg
parents: 75
diff changeset
139 }
80
6d97ef6bf7b7 Uploaded
greg
parents: 78
diff changeset
140 s_names = c(s_names, item);
77
8c4549f123da Uploaded
greg
parents: 75
diff changeset
141 }
61
5acfb82e4f44 Uploaded
greg
parents: 59
diff changeset
142 # Split state_colors into a list of strings.
80
6d97ef6bf7b7 Uploaded
greg
parents: 78
diff changeset
143 color_str = as.character(state_colors);
6d97ef6bf7b7 Uploaded
greg
parents: 78
diff changeset
144 items = strsplit(color_str, ",");
61
5acfb82e4f44 Uploaded
greg
parents: 59
diff changeset
145 for (item in items) {
80
6d97ef6bf7b7 Uploaded
greg
parents: 78
diff changeset
146 s_colors = c(s_colors, item);
61
5acfb82e4f44 Uploaded
greg
parents: 59
diff changeset
147 }
5acfb82e4f44 Uploaded
greg
parents: 59
diff changeset
148 }
77
8c4549f123da Uploaded
greg
parents: 75
diff changeset
149 base_track_file_name <- paste(tracks_dir, hub_name, sep="");
8c4549f123da Uploaded
greg
parents: 75
diff changeset
150 cells = create_cells(input_dir_state, chrom_len_file, base_track_file_name, s_indexes, s_names, s_colors);
83
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
151 cell_info = cbind(cells, cells, cells, "#000000");
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
152 cell_info = array(cell_info, dim=c(length(cells), 4));
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
153 cell_info = as.matrix(cell_info);
54
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
154 track_db = NULL;
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
155 for (i in 1:length(cells)) {
83
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
156 ii = which(cells[i] == cell_info[,1]);
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
157 if(length(ii) == 0) {
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
158 next;
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
159 }
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
160 ii = ii[1];
71
c04084f77b33 Uploaded
greg
parents: 70
diff changeset
161 # Get the color for the current state.
80
6d97ef6bf7b7 Uploaded
greg
parents: 78
diff changeset
162 color <- paste(c(col2rgb(s_colors[i])), collapse=",");
71
c04084f77b33 Uploaded
greg
parents: 70
diff changeset
163 # Get the bigDataUrl.
54
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
164 big_data_url <- get_big_data_url(galaxy_url, encoded_dataset_id, tracks_dir, i, build);
71
c04084f77b33 Uploaded
greg
parents: 70
diff changeset
165 # trackDb.txt track hub entry.
54
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
166 track_db = c(track_db, paste("track ", hub_name, "_track_", i, sep=""));
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
167 track_db = c(track_db, "type bigBed");
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
168 track_db = c(track_db, paste("bigDataUrl", big_data_url, sep=" "));
83
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
169 track_db = c(track_db, paste("shortLabel", cell_info[ii, 2], sep=" "));
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
170 track_db = c(track_db, paste("longLabel", paste(hub_name, cell_info[ii, 3], sep=" ")));
54
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
171 track_db = c(track_db, paste("priority", i));
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
172 track_db = c(track_db, "itemRgb on");
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
173 track_db = c(track_db, "maxItems 100000");
66
f0c488b73dbc Uploaded
greg
parents: 65
diff changeset
174 track_db = c(track_db, paste("color", color, sep=" "));
54
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
175 track_db = c(track_db, "visibility dense");
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
176 track_db = c(track_db, "");
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
177 }
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
178 return(track_db);
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
179 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
180
39
ecfe907615c5 Uploaded
greg
parents: 37
diff changeset
181 get_big_data_url = function(galaxy_url, encoded_dataset_id, tracks_dir, index, build) {
56
c54ef43eaee3 Uploaded
greg
parents: 55
diff changeset
182 track_files <- list.files(path=tracks_dir, pattern="\\.bigbed");
54
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
183 s <- paste(galaxy_url, 'datasets/', encoded_dataset_id, '/display/myHub/', build, '/', track_files[index], sep="");
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
184 return(s)
50
0c66347d5354 Uploaded
greg
parents: 48
diff changeset
185 }
0c66347d5354 Uploaded
greg
parents: 48
diff changeset
186
83
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
187 get_cell_type_names = function(state_file) {
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
188 # The first line of a state file is a comment
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
189 # that looks something like this:
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
190 # ID CHR POSst POSed E001 E002 PosClass
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
191 # The cell type names are the elemets whose
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
192 # 1-based indexes start at 5 and end with -1,
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
193 # which in the above case are E001 and E002.
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
194 fh = file(state_file,"r");
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
195 line = readLines(fh, n=1);
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
196 close(fh);
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
197 # Split line into a list of strings.
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
198 items = strsplit(line, "\\s+")[[1]];
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
199 # Extract the cell type names into a list.
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
200 last_cell_type_name_index = length(items) -1;
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
201 cell_type_names = c();
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
202 for (i in 5:last_cell_type_name_index) {
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
203 cell_type_names = c(cell_type_names, items[i]);
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
204 }
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
205 return(cell_type_names);
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
206 }
2f49b6bddccc Uploaded
greg
parents: 80
diff changeset
207
39
ecfe907615c5 Uploaded
greg
parents: 37
diff changeset
208 get_track_file_name = function(base_track_file_name, index, ext) {
54
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
209 track_file_name <- paste(base_track_file_name, index, ext, sep=".");
d7ab58fe039f Uploaded
greg
parents: 53
diff changeset
210 return(track_file_name);
39
ecfe907615c5 Uploaded
greg
parents: 37
diff changeset
211 }
ecfe907615c5 Uploaded
greg
parents: 37
diff changeset
212
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
213 # Create the hub.txt output.
53
8b1c49f2ecaa Uploaded
greg
parents: 52
diff changeset
214 hub_name_line <- paste("hub ", opt$hub_name, sep="");
77
8c4549f123da Uploaded
greg
parents: 75
diff changeset
215 hub_short_label_line <- paste("shortLabel ", opt$hub_short_label, sep="");
8c4549f123da Uploaded
greg
parents: 75
diff changeset
216 hub_long_label_line <- paste("longLabel ", opt$hub_long_label, sep="");
35
3e001c30b0f7 Uploaded
greg
parents: 34
diff changeset
217 genomes_txt_line <- paste("genomesFile genomes.txt", sep="");
39
ecfe907615c5 Uploaded
greg
parents: 37
diff changeset
218 email_line <- paste("email ", opt$email, sep="");
77
8c4549f123da Uploaded
greg
parents: 75
diff changeset
219 contents <- paste(hub_name_line, hub_short_label_line, hub_long_label_line, genomes_txt_line, email_line, sep="\n");
35
3e001c30b0f7 Uploaded
greg
parents: 34
diff changeset
220 hub_dir <- paste(opt$output_trackhub_files_path, "/", "myHub", "/", sep="");
22
a3740127fc5e Uploaded
greg
parents: 20
diff changeset
221 dir.create(hub_dir, showWarnings=FALSE);
35
3e001c30b0f7 Uploaded
greg
parents: 34
diff changeset
222 hub_file_path <- paste(hub_dir, "hub.txt", sep="");
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
223 write.table(contents, file=hub_file_path, quote=F, row.names=F, col.names=F);
696e435f5607 Uploaded
greg
parents: 18
diff changeset
224
696e435f5607 Uploaded
greg
parents: 18
diff changeset
225 # Create the genomes.txt output.
39
ecfe907615c5 Uploaded
greg
parents: 37
diff changeset
226 genome_line <- paste("genome ", opt$build, sep="");
35
3e001c30b0f7 Uploaded
greg
parents: 34
diff changeset
227 track_db_line <- paste("trackDb ", opt$build, "/", "trackDb.txt", sep="");
39
ecfe907615c5 Uploaded
greg
parents: 37
diff changeset
228 contents <- paste(genome_line, track_db_line, sep="\n");
30
4c42188478a7 Uploaded
greg
parents: 29
diff changeset
229 genomes_file_path <- paste(hub_dir, "genomes.txt", sep="");
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
230 write.table(contents, file=genomes_file_path, quote=F, row.names=F, col.names=F);
696e435f5607 Uploaded
greg
parents: 18
diff changeset
231
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
232 # Create the tracks.
68
06401685fb92 Uploaded
greg
parents: 66
diff changeset
233 heatmap_path <- paste(opt$script_dir, "create_heatmap.R", sep="/");
06401685fb92 Uploaded
greg
parents: 66
diff changeset
234 source(heatmap_path);
36
fe9dfd95f13d Uploaded
greg
parents: 35
diff changeset
235 tracks_dir <- paste(hub_dir, opt$build, "/", sep="");
11
45a2f84b8247 Uploaded
greg
parents: 9
diff changeset
236 dir.create(tracks_dir, showWarnings=FALSE);
70
fab0551a29ec Uploaded
greg
parents: 68
diff changeset
237 track_db <- create_track_db(opt$galaxy_url, opt$output_trackhub_id, opt$input_dir_para, opt$input_dir_state, opt$build,
77
8c4549f123da Uploaded
greg
parents: 75
diff changeset
238 opt$chrom_len_file, tracks_dir, opt$hub_name, opt$hub_short_label, opt$hub_long_label, opt$state_indexes,
8c4549f123da Uploaded
greg
parents: 75
diff changeset
239 opt$state_names, opt$state_colors);
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
240
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
241 # Create the trackDb.txt output.
35
3e001c30b0f7 Uploaded
greg
parents: 34
diff changeset
242 track_db_file_path <- paste(tracks_dir, "trackDb.txt", sep="");
13
f0dfff1128c8 Uploaded
greg
parents: 12
diff changeset
243 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
244
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
245 # Create the primary HTML dataset.
78
55da56869a0d Uploaded
greg
parents: 77
diff changeset
246 create_primary_html(opt$output_trackhub, tracks_dir, opt$build);