annotate ideas_genome_tracks.R @ 10:2b484c93d20a draft

Uploaded
author greg
date Thu, 16 Nov 2017 13:18:58 -0500
parents ecc2397d5aeb
children 45a2f84b8247
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"),
a842ab7e3bba Uploaded
greg
parents:
diff changeset
9 make_option(c("-d", "--header"), action="store", dest="header", default=NULL, help="Track header"),
a842ab7e3bba Uploaded
greg
parents:
diff changeset
10 make_option(c("-e", "--state_name"), action="store", dest="state_name", help="State name"),
a842ab7e3bba Uploaded
greg
parents:
diff changeset
11 make_option(c("-f", "--hub_id"), action="store", dest="hub_id", help="Not sure what this is"),
a842ab7e3bba Uploaded
greg
parents:
diff changeset
12 make_option(c("-i", "--cell_info"), action="store", dest="cell_info", default=NULL, help="Not sure what this is"),
a842ab7e3bba Uploaded
greg
parents:
diff changeset
13 make_option(c("-n", "--hub_name"), action="store", dest="hub_name", default=NULL, help="Not sure what this is"),
a842ab7e3bba Uploaded
greg
parents:
diff changeset
14 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
15 make_option(c("-q", "--input_dir_state"), action="store", dest="input_dir_state", help="Directory containing .state outputs from IDEAS"),
a842ab7e3bba Uploaded
greg
parents:
diff changeset
16 make_option(c("-u", "--output_track_db"), action="store", dest="output_track_db", help="Output track db file"),
a842ab7e3bba Uploaded
greg
parents:
diff changeset
17 make_option(c("-v", "--output_build"), action="store", dest="output_build", help="Output genome build file"),
9
ecc2397d5aeb Uploaded
greg
parents: 5
diff changeset
18 make_option(c("-w", "--output_hub"), action="store", dest="output_hub", help="Output hub file"),
ecc2397d5aeb Uploaded
greg
parents: 5
diff changeset
19 make_option(c("-x", "--output_hub_files_path"), action="store", dest="output_hub_files_path", help="Output hub extra files path")
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
20 )
a842ab7e3bba Uploaded
greg
parents:
diff changeset
21
a842ab7e3bba Uploaded
greg
parents:
diff changeset
22 parser <- OptionParser(usage="%prog [options] file", option_list=option_list)
a842ab7e3bba Uploaded
greg
parents:
diff changeset
23 args <- parse_args(parser, positional_arguments=TRUE)
a842ab7e3bba Uploaded
greg
parents:
diff changeset
24 opt <- args$options
a842ab7e3bba Uploaded
greg
parents:
diff changeset
25
a842ab7e3bba Uploaded
greg
parents:
diff changeset
26 create_color_scheme = function(input_dir_para) {
a842ab7e3bba Uploaded
greg
parents:
diff changeset
27 # Create the color scheme.
5
d1703a15c393 Uploaded
greg
parents: 0
diff changeset
28 para_files <- list.files(path=input_dir_para, full.names=TRUE);
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
29 mc = NULL;
5
d1703a15c393 Uploaded
greg
parents: 0
diff changeset
30 x = read.table(para_files[1], comment="!", nrows=1);
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
31 l = as.integer(regexpr("\\*", as.matrix(x)));
a842ab7e3bba Uploaded
greg
parents:
diff changeset
32 l = min(which(l>0)) - 2;
5
d1703a15c393 Uploaded
greg
parents: 0
diff changeset
33 x = as.matrix(read.table(para_files[1]));
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
34 if (length(para_files) > 1) {
a842ab7e3bba Uploaded
greg
parents:
diff changeset
35 for (i in 2:length(para_files)) {
5
d1703a15c393 Uploaded
greg
parents: 0
diff changeset
36 x = x + as.matrix(read.table(para_files[i]));
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
37 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
38 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
39 p = x[,1] / sum(x[,1]);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
40 m = array(as.matrix(x[,1:l+1] / x[,1]), dim=c(dim(x)[1], l));
5
d1703a15c393 Uploaded
greg
parents: 0
diff changeset
41 state_color <- get_state_color(m, mc);
d1703a15c393 Uploaded
greg
parents: 0
diff changeset
42 state_color
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
43 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
44
a842ab7e3bba Uploaded
greg
parents:
diff changeset
45 get_state_color = function(statemean, markcolor=NULL)
a842ab7e3bba Uploaded
greg
parents:
diff changeset
46 {
a842ab7e3bba Uploaded
greg
parents:
diff changeset
47 if (length(markcolor) == 0) {
a842ab7e3bba Uploaded
greg
parents:
diff changeset
48 markcolor = rep("", dim(statemean)[2]);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
49 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
50 markcolor = t(col2rgb(markcolor));
a842ab7e3bba Uploaded
greg
parents:
diff changeset
51 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
52 rg = apply(statemean, 1, range);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
53 mm = NULL;
a842ab7e3bba Uploaded
greg
parents:
diff changeset
54 for(i in 1:dim(statemean)[1]) {
a842ab7e3bba Uploaded
greg
parents:
diff changeset
55 mm = rbind(mm, (statemean[i,]-rg[1,i])/(rg[2,i]-rg[1,i]+1e-10));
a842ab7e3bba Uploaded
greg
parents:
diff changeset
56 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
57 mm = mm^6;
a842ab7e3bba Uploaded
greg
parents:
diff changeset
58 mm = mm / (apply(mm, 1, sum) + 1e-10);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
59 mycol = mm%*%markcolor;
a842ab7e3bba Uploaded
greg
parents:
diff changeset
60 s = apply(statemean, 1, max);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
61 s = (s - min(s)) / (max(s) - min(s) + 1e-10);
5
d1703a15c393 Uploaded
greg
parents: 0
diff changeset
62 h = t(apply(mycol, 1, function(x) {rgb2hsv(x[1], x[2], x[3])}));
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
63 h[,2] = h[,2] * s;
5
d1703a15c393 Uploaded
greg
parents: 0
diff changeset
64 h = apply(h, 1, function(x) {hsv(x[1], x[2], x[3])});
d1703a15c393 Uploaded
greg
parents: 0
diff changeset
65 rt = cbind(apply(t(col2rgb(h)), 1, function(x) {paste(x, collapse=",")}), h);
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
66 return(rt);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
67 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
68
a842ab7e3bba Uploaded
greg
parents:
diff changeset
69 create_track = function(input_dir_state, chrom_len_file, outpref, state_color, header, state_name) {
5
d1703a15c393 Uploaded
greg
parents: 0
diff changeset
70 state_files <- list.files(path=input_dir_state, full.names=TRUE);
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
71 genome_size = read.table(chrom_len_file);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
72 g = NULL;
a842ab7e3bba Uploaded
greg
parents:
diff changeset
73 for(i in state_files) {
a842ab7e3bba Uploaded
greg
parents:
diff changeset
74 tg = as.matrix(fread(i));
a842ab7e3bba Uploaded
greg
parents:
diff changeset
75 t = NULL;
a842ab7e3bba Uploaded
greg
parents:
diff changeset
76 for(j in 1:dim(genome_size)[1]) {
a842ab7e3bba Uploaded
greg
parents:
diff changeset
77 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
78 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
79 if (length(t) > 0) {
a842ab7e3bba Uploaded
greg
parents:
diff changeset
80 tg = tg[-t,];
a842ab7e3bba Uploaded
greg
parents:
diff changeset
81 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
82 t = which(is.na(match(tg[,2], genome_size[,1]))==T);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
83 if (length(t)>0) {
a842ab7e3bba Uploaded
greg
parents:
diff changeset
84 tg = tg[-t,];
a842ab7e3bba Uploaded
greg
parents:
diff changeset
85 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
86 g = rbind(g, tg);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
87 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
88 uchr = sort(unique(as.character(g[,2])));
a842ab7e3bba Uploaded
greg
parents:
diff changeset
89 g1 = NULL;
5
d1703a15c393 Uploaded
greg
parents: 0
diff changeset
90 for(i in uchr) {
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
91 t = which(g[,2]==i);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
92 g1 = rbind(g1, g[t[order(as.integer(g[t, 3]))],]);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
93 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
94 g = NULL;
a842ab7e3bba Uploaded
greg
parents:
diff changeset
95 chr = as.character(g1[,2]);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
96 posst = as.numeric(g1[,3]);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
97 posed = as.numeric(g1[,4]);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
98 state = as.matrix(g1[,5:(dim(g1)[2]-1)]);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
99 if (length(state_name)==0) {
a842ab7e3bba Uploaded
greg
parents:
diff changeset
100 state_name=0:max(state);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
101 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
102 L = dim(g1)[1];
a842ab7e3bba Uploaded
greg
parents:
diff changeset
103 n = dim(state)[2];
a842ab7e3bba Uploaded
greg
parents:
diff changeset
104 if (length(header) > 0) {
a842ab7e3bba Uploaded
greg
parents:
diff changeset
105 colnames(g1) = header;
a842ab7e3bba Uploaded
greg
parents:
diff changeset
106 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
107 cells = as.character(colnames(g1)[5:(dim(g1)[2]-1)]);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
108 g1 = NULL;
a842ab7e3bba Uploaded
greg
parents:
diff changeset
109 options(scipen=999);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
110 tt = which(chr[2:L]!=chr[2:L-1]);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
111 tt = c(tt,which(posst[2:L]!=posed[2:L-1]));
a842ab7e3bba Uploaded
greg
parents:
diff changeset
112 tt = sort(unique(tt));
a842ab7e3bba Uploaded
greg
parents:
diff changeset
113 for(i in 1:n) {
a842ab7e3bba Uploaded
greg
parents:
diff changeset
114 tstate = state[,i];
5
d1703a15c393 Uploaded
greg
parents: 0
diff changeset
115 t = c(tt, which(tstate[2:L]!=tstate[2:L-1]));
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
116 t = sort(unique(t));
a842ab7e3bba Uploaded
greg
parents:
diff changeset
117 t0 = c(0, t) + 1;
a842ab7e3bba Uploaded
greg
parents:
diff changeset
118 t = c(t, L);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
119 np = cbind(chr[t], posst[t0], posed[t], tstate[t]);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
120 x = cbind(np[,1:3], state_name[as.integer(np[,4])+1], 1000, ".", np[,2:3], state_color[as.numeric(np[,4])+1]);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
121 write.table(as.matrix(x), paste(outpref, i, "bed1", sep="."), quote=F, row.names=F, col.names=F);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
122 system(paste("bedToBigBed ", outpref, ".", i, ".bed1 ", chrom_len_file, " ", outpref, ".", i, ".bb", sep=""));
a842ab7e3bba Uploaded
greg
parents:
diff changeset
123 system(paste("rm ", paste(outpref, i, "bed1", sep=".")));
a842ab7e3bba Uploaded
greg
parents:
diff changeset
124 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
125 return(cells);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
126 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
127
a842ab7e3bba Uploaded
greg
parents:
diff changeset
128 create_track_db = function(input_dir_state, chrom_len_file, track_file_name, state_color, header, state_name, hub_name, cell_info) {
a842ab7e3bba Uploaded
greg
parents:
diff changeset
129 cells = create_track(input_dir_state, chrom_len_file, track_file_name, state_color, header, state_name);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
130 if (length(cell_info) == 0) {
a842ab7e3bba Uploaded
greg
parents:
diff changeset
131 cell_info = cbind(cells, cells, cells, "#000000");
a842ab7e3bba Uploaded
greg
parents:
diff changeset
132 cell_info = array(cell_info, dim=c(length(cells), 4));
a842ab7e3bba Uploaded
greg
parents:
diff changeset
133 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
134 cell_info = as.matrix(cell_info);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
135 track_db = NULL;
a842ab7e3bba Uploaded
greg
parents:
diff changeset
136 for(i in 1:length(cells)) {
a842ab7e3bba Uploaded
greg
parents:
diff changeset
137 ii = which(cells[i] == cell_info[,1]);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
138 if (length(ii) == 0) {
a842ab7e3bba Uploaded
greg
parents:
diff changeset
139 next;
a842ab7e3bba Uploaded
greg
parents:
diff changeset
140 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
141 ii = ii[1];
a842ab7e3bba Uploaded
greg
parents:
diff changeset
142 track_db = c(track_db, paste("track bigBed", i, sep=""));
a842ab7e3bba Uploaded
greg
parents:
diff changeset
143 track_db = c(track_db, paste("priority", ii));
a842ab7e3bba Uploaded
greg
parents:
diff changeset
144 track_db = c(track_db, "type bigBed 9 .");
a842ab7e3bba Uploaded
greg
parents:
diff changeset
145 track_db = c(track_db, "itemRgb on");
a842ab7e3bba Uploaded
greg
parents:
diff changeset
146 track_db = c(track_db, "maxItems 100000");
a842ab7e3bba Uploaded
greg
parents:
diff changeset
147 track_db = c(track_db, paste("shortLabel", cell_info[ii, 2]));
a842ab7e3bba Uploaded
greg
parents:
diff changeset
148 track_db = c(track_db, paste("longLabel", paste(hub_name, cell_info[ii, 3])));
a842ab7e3bba Uploaded
greg
parents:
diff changeset
149 track_db = c(track_db, paste("color", paste(c(col2rgb(cell_info[ii, 4])), collapse=",")));
a842ab7e3bba Uploaded
greg
parents:
diff changeset
150 track_db = c(track_db, "visibility dense");
a842ab7e3bba Uploaded
greg
parents:
diff changeset
151 track_db = c(track_db, "");
a842ab7e3bba Uploaded
greg
parents:
diff changeset
152 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
153 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
154
a842ab7e3bba Uploaded
greg
parents:
diff changeset
155 if (length(opt$hub_name) == 0) {
9
ecc2397d5aeb Uploaded
greg
parents: 5
diff changeset
156 hub_name <- opt$hub_id;
ecc2397d5aeb Uploaded
greg
parents: 5
diff changeset
157 } else {
ecc2397d5aeb Uploaded
greg
parents: 5
diff changeset
158 hub_name <- opt$hub_name;
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
159 }
a842ab7e3bba Uploaded
greg
parents:
diff changeset
160
a842ab7e3bba Uploaded
greg
parents:
diff changeset
161 # Create the tracks output directory.
a842ab7e3bba Uploaded
greg
parents:
diff changeset
162 tracks_output_dir = paste("tracks_", opt$hub_id, "/", sep="");
a842ab7e3bba Uploaded
greg
parents:
diff changeset
163 dir.create(tracks_output_dir, showWarnings=FALSE);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
164
a842ab7e3bba Uploaded
greg
parents:
diff changeset
165 # Create the color scheme.
a842ab7e3bba Uploaded
greg
parents:
diff changeset
166 state_color <- create_color_scheme(opt$input_dir_para);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
167
a842ab7e3bba Uploaded
greg
parents:
diff changeset
168 # Create the tracks.
9
ecc2397d5aeb Uploaded
greg
parents: 5
diff changeset
169 track_db <- create_track_db(opt$input_dir_state, opt$chrom_len_file, paste(tracks_output_dir, opt$hub_id, sep=""), state_color, opt$header, opt$state_name, hub_name, opt$cell_info);
0
a842ab7e3bba Uploaded
greg
parents:
diff changeset
170
a842ab7e3bba Uploaded
greg
parents:
diff changeset
171 # Write the outputs.
a842ab7e3bba Uploaded
greg
parents:
diff changeset
172 write.table(track_db, opt$output_track_db, quote=F, row.names=F, col.names=F);
a842ab7e3bba Uploaded
greg
parents:
diff changeset
173 write.table(c(paste("genome", opt$build), opt$output_build), paste(tracks_output_dir, "genomes_", opt$hub_id, ".txt", sep=""), quote=F, row.names=F, col.names=F);
9
ecc2397d5aeb Uploaded
greg
parents: 5
diff changeset
174 write.table(c(paste("hub", opt$hub_id), paste("shortLabel", opt$hub_id), paste("longLabel", hub_name), paste("genomesFile genomes_", opt$hub_id, ".txt", sep=""), "email yzz2 at psu.edu"), paste(opt$output_hub_files_path, "hub_", opt$hub_id, ".txt", sep=""), quote=F, row.names=F, col.names=F);