annotate ideas_genome_tracks.R @ 20:696e435f5607 draft

Uploaded
author greg
date Mon, 20 Nov 2017 10:34:37 -0500
parents 0bd424ce10fb
children a3740127fc5e
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(
a842ab7e3bba Uploaded
greg
parents:
diff changeset
7 make_option(c("-b", "--build"), action="store", dest="build", help="Genome build"),
a842ab7e3bba Uploaded
greg
parents:
diff changeset
8 make_option(c("-c", "--chrom_len_file"), action="store", dest="chrom_len_file", help="Chromosome length file"),
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
9 make_option(c("-e", "--email"), action="store", dest="email", help="User email address"),
696e435f5607 Uploaded
greg
parents: 18
diff changeset
10 make_option(c("-l", "--long_label"), action="store", dest="long_label", help="Hub long label"),
696e435f5607 Uploaded
greg
parents: 18
diff changeset
11 make_option(c("-n", "--hub_name"), action="store", dest="hub_name", default=NULL, help="Hub name without spaces"),
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
12 make_option(c("-p", "--input_dir_para"), action="store", dest="input_dir_para", help="Directory containing .para outputs from IDEAS"),
a842ab7e3bba Uploaded
greg
parents:
diff changeset
13 make_option(c("-q", "--input_dir_state"), action="store", dest="input_dir_state", help="Directory containing .state outputs from IDEAS"),
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
14 make_option(c("-s", "--short_label"), action="store", dest="short_label", help="Hub short label"),
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
15 make_option(c("-u", "--output_track_db"), action="store", dest="output_track_db", help="Output track db file"),
11
45a2f84b8247 Uploaded
greg
parents: 9
diff changeset
16 make_option(c("-w", "--output_trackhub"), action="store", dest="output_trackhub", help="Output hub file"),
45a2f84b8247 Uploaded
greg
parents: 9
diff changeset
17 make_option(c("-x", "--output_trackhub_files_path"), action="store", dest="output_trackhub_files_path", help="Output hub extra files path")
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
18 )
a842ab7e3bba Uploaded
greg
parents:
diff changeset
19
a842ab7e3bba Uploaded
greg
parents:
diff changeset
20 parser <- OptionParser(usage="%prog [options] file", option_list=option_list)
a842ab7e3bba Uploaded
greg
parents:
diff changeset
21 args <- parse_args(parser, positional_arguments=TRUE)
a842ab7e3bba Uploaded
greg
parents:
diff changeset
22 opt <- args$options
a842ab7e3bba Uploaded
greg
parents:
diff changeset
23
a842ab7e3bba Uploaded
greg
parents:
diff changeset
24 create_color_scheme = function(input_dir_para) {
a842ab7e3bba Uploaded
greg
parents:
diff changeset
25 # Create the color scheme.
5
d1703a15c393 Uploaded
greg
parents: 0
diff changeset
26 para_files <- list.files(path=input_dir_para, full.names=TRUE);
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
27 mc = NULL;
5
d1703a15c393 Uploaded
greg
parents: 0
diff changeset
28 x = read.table(para_files[1], comment="!", nrows=1);
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
29 l = as.integer(regexpr("\\*", as.matrix(x)));
a842ab7e3bba Uploaded
greg
parents:
diff changeset
30 l = min(which(l>0)) - 2;
5
d1703a15c393 Uploaded
greg
parents: 0
diff changeset
31 x = as.matrix(read.table(para_files[1]));
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
32 if (length(para_files) > 1) {
a842ab7e3bba Uploaded
greg
parents:
diff changeset
33 for (i in 2:length(para_files)) {
5
d1703a15c393 Uploaded
greg
parents: 0
diff changeset
34 x = x + as.matrix(read.table(para_files[i]));
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
35 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
36 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
37 p = x[,1] / sum(x[,1]);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
38 m = array(as.matrix(x[,1:l+1] / x[,1]), dim=c(dim(x)[1], l));
5
d1703a15c393 Uploaded
greg
parents: 0
diff changeset
39 state_color <- get_state_color(m, mc);
11
45a2f84b8247 Uploaded
greg
parents: 9
diff changeset
40 return(state_color);
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
41 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
42
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
43 get_track_file_name = function(base_track_file_name, index, ext) {
696e435f5607 Uploaded
greg
parents: 18
diff changeset
44 track_file_name <- paste(base_track_file_name, index, ext, sep=".");
696e435f5607 Uploaded
greg
parents: 18
diff changeset
45 return(track_file_name);
696e435f5607 Uploaded
greg
parents: 18
diff changeset
46 }
696e435f5607 Uploaded
greg
parents: 18
diff changeset
47
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
48 get_state_color = function(statemean, markcolor=NULL)
a842ab7e3bba Uploaded
greg
parents:
diff changeset
49 {
a842ab7e3bba Uploaded
greg
parents:
diff changeset
50 if (length(markcolor) == 0) {
a842ab7e3bba Uploaded
greg
parents:
diff changeset
51 markcolor = rep("", dim(statemean)[2]);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
52 markcolor[order(apply(statemean, 2, sd), decreasing=T)] = hsv((1:dim(statemean)[2]-1)/dim(statemean)[2], 1, 1);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
53 markcolor = t(col2rgb(markcolor));
a842ab7e3bba Uploaded
greg
parents:
diff changeset
54 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
55 rg = apply(statemean, 1, range);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
56 mm = NULL;
a842ab7e3bba Uploaded
greg
parents:
diff changeset
57 for(i in 1:dim(statemean)[1]) {
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
58 mm = rbind(mm, (statemean[i,]-rg[1,i])+1e-10)/(rg[2,i]-rg[1,i]+1e-10));
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
59 }
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
60 mm = mm^6;
696e435f5607 Uploaded
greg
parents: 18
diff changeset
61 if(dim(mm)[2] > 1) {
696e435f5607 Uploaded
greg
parents: 18
diff changeset
62 mm = mm / (apply(mm, 1, sum) + 1e-10);
696e435f5607 Uploaded
greg
parents: 18
diff changeset
63 }
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
64 mycol = mm%*%markcolor;
a842ab7e3bba Uploaded
greg
parents:
diff changeset
65 s = apply(statemean, 1, max);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
66 s = (s - min(s)) / (max(s) - min(s) + 1e-10);
5
d1703a15c393 Uploaded
greg
parents: 0
diff changeset
67 h = t(apply(mycol, 1, function(x) {rgb2hsv(x[1], x[2], x[3])}));
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
68 h[,2] = h[,2] * s;
5
d1703a15c393 Uploaded
greg
parents: 0
diff changeset
69 h = apply(h, 1, function(x) {hsv(x[1], x[2], x[3])});
d1703a15c393 Uploaded
greg
parents: 0
diff changeset
70 rt = cbind(apply(t(col2rgb(h)), 1, function(x) {paste(x, collapse=",")}), h);
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
71 return(rt);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
72 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
73
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
74 create_primary_html = function(output_trackhub, tracks_dir) {
696e435f5607 Uploaded
greg
parents: 18
diff changeset
75 track_files <- list.files(path=tracks_dir, full.names=TRUE);
696e435f5607 Uploaded
greg
parents: 18
diff changeset
76 s <- "<html><head></head><body><ul>"
696e435f5607 Uploaded
greg
parents: 18
diff changeset
77 for (i in 1:length(track_files)) {
696e435f5607 Uploaded
greg
parents: 18
diff changeset
78 track_file <- track_files[i];
696e435f5607 Uploaded
greg
parents: 18
diff changeset
79 s <- paste(s, "<li><a href=\"", track_file, "\">", track_file, "</a></li>");
696e435f5607 Uploaded
greg
parents: 18
diff changeset
80 s <- paste(s, "</ul></body></html>"
696e435f5607 Uploaded
greg
parents: 18
diff changeset
81 sink(output_trackhub);
696e435f5607 Uploaded
greg
parents: 18
diff changeset
82 cat(s);
696e435f5607 Uploaded
greg
parents: 18
diff changeset
83 sink();
696e435f5607 Uploaded
greg
parents: 18
diff changeset
84 }
696e435f5607 Uploaded
greg
parents: 18
diff changeset
85
696e435f5607 Uploaded
greg
parents: 18
diff changeset
86 create_track = function(input_dir_state, chrom_len_file, base_track_file_name, state_color) {
5
d1703a15c393 Uploaded
greg
parents: 0
diff changeset
87 state_files <- list.files(path=input_dir_state, full.names=TRUE);
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
88 genome_size = read.table(chrom_len_file);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
89 g = NULL;
a842ab7e3bba Uploaded
greg
parents:
diff changeset
90 for(i in state_files) {
a842ab7e3bba Uploaded
greg
parents:
diff changeset
91 tg = as.matrix(fread(i));
a842ab7e3bba Uploaded
greg
parents:
diff changeset
92 t = NULL;
a842ab7e3bba Uploaded
greg
parents:
diff changeset
93 for(j in 1:dim(genome_size)[1]) {
a842ab7e3bba Uploaded
greg
parents:
diff changeset
94 t = c(t, which(tg[,2]==as.character(genome_size[j, 1]) & as.numeric(tg[,4]) > as.numeric(genome_size[j, 2])));
a842ab7e3bba Uploaded
greg
parents:
diff changeset
95 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
96 if (length(t) > 0) {
a842ab7e3bba Uploaded
greg
parents:
diff changeset
97 tg = tg[-t,];
a842ab7e3bba Uploaded
greg
parents:
diff changeset
98 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
99 t = which(is.na(match(tg[,2], genome_size[,1]))==T);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
100 if (length(t)>0) {
a842ab7e3bba Uploaded
greg
parents:
diff changeset
101 tg = tg[-t,];
a842ab7e3bba Uploaded
greg
parents:
diff changeset
102 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
103 g = rbind(g, tg);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
104 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
105 uchr = sort(unique(as.character(g[,2])));
a842ab7e3bba Uploaded
greg
parents:
diff changeset
106 g1 = NULL;
5
d1703a15c393 Uploaded
greg
parents: 0
diff changeset
107 for(i in uchr) {
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
108 t = which(g[,2]==i);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
109 g1 = rbind(g1, g[t[order(as.integer(g[t, 3]))],]);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
110 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
111 g = NULL;
a842ab7e3bba Uploaded
greg
parents:
diff changeset
112 chr = as.character(g1[,2]);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
113 posst = as.numeric(g1[,3]);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
114 posed = as.numeric(g1[,4]);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
115 state = as.matrix(g1[,5:(dim(g1)[2]-1)]);
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
116 state_name = 0:max(state);
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
117 L = dim(g1)[1];
a842ab7e3bba Uploaded
greg
parents:
diff changeset
118 n = dim(state)[2];
a842ab7e3bba Uploaded
greg
parents:
diff changeset
119 cells = as.character(colnames(g1)[5:(dim(g1)[2]-1)]);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
120 g1 = NULL;
a842ab7e3bba Uploaded
greg
parents:
diff changeset
121 options(scipen=999);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
122 tt = which(chr[2:L]!=chr[2:L-1]);
11
45a2f84b8247 Uploaded
greg
parents: 9
diff changeset
123 tt = c(tt, which(posst[2:L]!=posed[2:L-1]));
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
124 tt = sort(unique(tt));
a842ab7e3bba Uploaded
greg
parents:
diff changeset
125 for(i in 1:n) {
a842ab7e3bba Uploaded
greg
parents:
diff changeset
126 tstate = state[,i];
5
d1703a15c393 Uploaded
greg
parents: 0
diff changeset
127 t = c(tt, which(tstate[2:L]!=tstate[2:L-1]));
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
128 t = sort(unique(t));
a842ab7e3bba Uploaded
greg
parents:
diff changeset
129 t0 = c(0, t) + 1;
a842ab7e3bba Uploaded
greg
parents:
diff changeset
130 t = c(t, L);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
131 np = cbind(chr[t], posst[t0], posed[t], tstate[t]);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
132 x = cbind(np[,1:3], state_name[as.integer(np[,4])+1], 1000, ".", np[,2:3], state_color[as.numeric(np[,4])+1]);
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
133 track_file_name_bed <- get_track_file_name(base_track_file_name, i, "bed");
696e435f5607 Uploaded
greg
parents: 18
diff changeset
134 track_file_name_bigbed <- get_track_file_name(base_track_file_name, i, "bigbed");
11
45a2f84b8247 Uploaded
greg
parents: 9
diff changeset
135 write.table(as.matrix(x), track_file_name_bed, quote=F, row.names=F, col.names=F);
45a2f84b8247 Uploaded
greg
parents: 9
diff changeset
136 system(paste("bedToBigBed ", track_file_name_bed, chrom_len_file, " ", track_file_name_bigbed));
15
f8c479d219bd Uploaded
greg
parents: 14
diff changeset
137 system(paste("rm ", track_file_name_bed));
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
138 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
139 return(cells);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
140 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
141
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
142 create_track_db = function(input_dir_state, chrom_len_file, tracks_dir, hub_name, short_label, long_label, state_color) {
696e435f5607 Uploaded
greg
parents: 18
diff changeset
143 base_track_file_name <- paste(tracks_dir, hub_name, sep="");
696e435f5607 Uploaded
greg
parents: 18
diff changeset
144 cells = create_track(input_dir_state, chrom_len_file, base_track_file_name, state_color);
696e435f5607 Uploaded
greg
parents: 18
diff changeset
145 cell_info = cbind(cells, cells, cells, "#000000");
696e435f5607 Uploaded
greg
parents: 18
diff changeset
146 cell_info = array(cell_info, dim=c(length(cells), 4));
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
147 cell_info = as.matrix(cell_info);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
148 track_db = NULL;
a842ab7e3bba Uploaded
greg
parents:
diff changeset
149 for(i in 1:length(cells)) {
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
150
696e435f5607 Uploaded
greg
parents: 18
diff changeset
151 ii = which(cells[i] == cellinfo[,1]);
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
152 if (length(ii) == 0) {
a842ab7e3bba Uploaded
greg
parents:
diff changeset
153 next;
a842ab7e3bba Uploaded
greg
parents:
diff changeset
154 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
155 ii = ii[1];
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
156 # trackDb.txt track entry.
696e435f5607 Uploaded
greg
parents: 18
diff changeset
157 track_db = c(track_db, paste(hub_name, "_track_", i, sep=""));
696e435f5607 Uploaded
greg
parents: 18
diff changeset
158 track_db = c(track_db, "type bigBed");
696e435f5607 Uploaded
greg
parents: 18
diff changeset
159 track_db = c(track_db, paste("bigDataUrl ", get_track_file_name(base_track_file_name, i, "bigbed"));
696e435f5607 Uploaded
greg
parents: 18
diff changeset
160 track_db = c(track_db, paste("shortLabel", short_label));
696e435f5607 Uploaded
greg
parents: 18
diff changeset
161 track_db = c(track_db, paste("longLabel", long_label);
696e435f5607 Uploaded
greg
parents: 18
diff changeset
162 track_db = c(track_db, paste("priority", i));
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
163 track_db = c(track_db, "itemRgb on");
a842ab7e3bba Uploaded
greg
parents:
diff changeset
164 track_db = c(track_db, "maxItems 100000");
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
165 track_db = c(track_db, paste("color", paste(c(col2rgb(cellinfo[ii,4])), collapse=",")));
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
166 track_db = c(track_db, "visibility dense");
a842ab7e3bba Uploaded
greg
parents:
diff changeset
167 track_db = c(track_db, "");
a842ab7e3bba Uploaded
greg
parents:
diff changeset
168 }
11
45a2f84b8247 Uploaded
greg
parents: 9
diff changeset
169 return(track_db);
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
170 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
171
a842ab7e3bba Uploaded
greg
parents:
diff changeset
172 # Create the color scheme.
a842ab7e3bba Uploaded
greg
parents:
diff changeset
173 state_color <- create_color_scheme(opt$input_dir_para);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
174
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
175 # Create the hub.txt output.
696e435f5607 Uploaded
greg
parents: 18
diff changeset
176 contents <- c(paste("hub", opt$hub_name), paste("shortLabel",opt$short_label), paste("longLabel", opt$long_label), paste("genomesFile genomes.txt", sep=""), paste("email", opt$email))
696e435f5607 Uploaded
greg
parents: 18
diff changeset
177 hub_file_path <- paste(opt$output_trackhub_files_path, "/", "hub.txt", sep="")
696e435f5607 Uploaded
greg
parents: 18
diff changeset
178 write.table(contents, file=hub_file_path, quote=F, row.names=F, col.names=F);
696e435f5607 Uploaded
greg
parents: 18
diff changeset
179
696e435f5607 Uploaded
greg
parents: 18
diff changeset
180 # Create the genomes.txt output.
696e435f5607 Uploaded
greg
parents: 18
diff changeset
181 contents <- c(paste("genome", opt$build), paste("trackDb ", opt$build, "/", "trackDb.txt", sep=""))
696e435f5607 Uploaded
greg
parents: 18
diff changeset
182 genomes_file_path <- paste(opt$output_trackhub_files_path, "/", "genomes.txt", sep="")
696e435f5607 Uploaded
greg
parents: 18
diff changeset
183 write.table(contents, file=genomes_file_path, quote=F, row.names=F, col.names=F);
696e435f5607 Uploaded
greg
parents: 18
diff changeset
184
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
185 # Create the tracks.
18
0bd424ce10fb Uploaded
greg
parents: 15
diff changeset
186 tracks_dir <- paste(opt$output_trackhub_files_path, "/", "tracks", "/", sep="");
11
45a2f84b8247 Uploaded
greg
parents: 9
diff changeset
187 dir.create(tracks_dir, showWarnings=FALSE);
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
188 track_db <- create_track_db(opt$input_dir_state, opt$chrom_len_file, tracks_dir, opt$hub_name, opt$short_label, opt$long_label, state_color);
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
189
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
190 # Create the trackDb.txt output.
15
f8c479d219bd Uploaded
greg
parents: 14
diff changeset
191 track_db_file_path <- paste(tracks_dir, "/", paste("trackDb.txt", sep=""), sep="")
13
f0dfff1128c8 Uploaded
greg
parents: 12
diff changeset
192 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
193
20
696e435f5607 Uploaded
greg
parents: 18
diff changeset
194 # Create the primary HTML dataset.
696e435f5607 Uploaded
greg
parents: 18
diff changeset
195 create_primary_html(opt$output_trackhub, tracks_dir);