Mercurial > repos > greg > ideas_genome_tracks

changeset 11:45a2f84b8247 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author greg
date Thu, 16 Nov 2017 13:20:13 -0500
parents 2b484c93d20a
children 0f385fb474ca
files ideas_genome_tracks.R
diffstat 1 files changed, 28 insertions(+), 19 deletions(-) [+]
line wrap: on
line diff
--- a/ideas_genome_tracks.R	Thu Nov 16 13:18:58 2017 -0500
+++ b/ideas_genome_tracks.R	Thu Nov 16 13:20:13 2017 -0500
@@ -9,14 +9,14 @@
     make_option(c("-d", "--header"),  action="store", dest="header", default=NULL, help="Track header"),
     make_option(c("-e", "--state_name"),  action="store", dest="state_name", help="State name"),
     make_option(c("-f", "--hub_id"),  action="store", dest="hub_id", help="Not sure what this is"),
+    make_option(c("-g", "--email"),  action="store", dest="email", help="User email address"),
     make_option(c("-i", "--cell_info"),  action="store", dest="cell_info", default=NULL, help="Not sure what this is"),
     make_option(c("-n", "--hub_name"),  action="store", dest="hub_name", default=NULL, help="Not sure what this is"),
     make_option(c("-p", "--input_dir_para"), action="store", dest="input_dir_para", help="Directory containing .para outputs from IDEAS"),
     make_option(c("-q", "--input_dir_state"), action="store", dest="input_dir_state", help="Directory containing .state outputs from IDEAS"),
     make_option(c("-u", "--output_track_db"),  action="store", dest="output_track_db", help="Output track db file"),
-    make_option(c("-v", "--output_build"),  action="store", dest="output_build", help="Output genome build file"),
-    make_option(c("-w", "--output_hub"),  action="store", dest="output_hub", help="Output hub file"),
-    make_option(c("-x", "--output_hub_files_path"),  action="store", dest="output_hub_files_path", help="Output hub extra files path")
+    make_option(c("-w", "--output_trackhub"),  action="store", dest="output_trackhub", help="Output hub file"),
+    make_option(c("-x", "--output_trackhub_files_path"),  action="store", dest="output_trackhub_files_path", help="Output hub extra files path")
 )
 
 parser <- OptionParser(usage="%prog [options] file", option_list=option_list)
@@ -39,7 +39,7 @@
     p = x[,1] / sum(x[,1]);
     m = array(as.matrix(x[,1:l+1] / x[,1]), dim=c(dim(x)[1], l));
     state_color <- get_state_color(m, mc);
-    state_color
+    return(state_color);
 }
 
 get_state_color = function(statemean, markcolor=NULL)
@@ -66,7 +66,7 @@
     return(rt);
 }
 
-create_track = function(input_dir_state, chrom_len_file, outpref, state_color, header, state_name) {
+create_track = function(input_dir_state, chrom_len_file, base_track_file_name, state_color, state_name, header) {
     state_files <- list.files(path=input_dir_state, full.names=TRUE);
     genome_size = read.table(chrom_len_file);
     g = NULL;
@@ -108,7 +108,7 @@
     g1 = NULL;
     options(scipen=999);
     tt = which(chr[2:L]!=chr[2:L-1]);
-    tt = c(tt,which(posst[2:L]!=posed[2:L-1]));
+    tt = c(tt, which(posst[2:L]!=posed[2:L-1]));
     tt = sort(unique(tt));
     for(i in 1:n) {
         tstate = state[,i];
@@ -118,15 +118,18 @@
         t = c(t, L);
         np = cbind(chr[t], posst[t0], posed[t], tstate[t]);
         x = cbind(np[,1:3], state_name[as.integer(np[,4])+1], 1000, ".", np[,2:3], state_color[as.numeric(np[,4])+1]);
-        write.table(as.matrix(x), paste(outpref, i, "bed1", sep="."), quote=F, row.names=F, col.names=F);
-        system(paste("bedToBigBed ", outpref, ".", i, ".bed1 ", chrom_len_file, " ", outpref, ".", i, ".bb", sep=""));
-        system(paste("rm ", paste(outpref, i, "bed1", sep=".")));
+        track_file_name_bed <- paste(base_track_file_name, i, "bed", sep=".")
+        track_file_name_bigbed <- paste(base_track_file_name, i, "bigbed", sep=".")
+        write.table(as.matrix(x), track_file_name_bed, quote=F, row.names=F, col.names=F);
+        system(paste("bedToBigBed ", track_file_name_bed, chrom_len_file, " ", track_file_name_bigbed));
+        #system(paste("rm ", track_file_name_bed));
     }
     return(cells);
 }
 
-create_track_db = function(input_dir_state, chrom_len_file, track_file_name, state_color, header, state_name, hub_name, cell_info) {
-    cells = create_track(input_dir_state, chrom_len_file, track_file_name, state_color, header, state_name);
+create_track_db = function(input_dir_state, chrom_len_file, header, tracks_dir, hub_id, hub_name, state_color, state_name, cell_info) {
+    base_track_file_name <- paste(tracks_dir, hub_id, sep="");
+    cells = create_track(input_dir_state, chrom_len_file, base_track_file_name, state_color, state_name, header);
     if (length(cell_info) == 0) {
         cell_info = cbind(cells, cells, cells, "#000000");
         cell_info = array(cell_info, dim=c(length(cells), 4));
@@ -150,6 +153,7 @@
         track_db = c(track_db, "visibility dense");
         track_db = c(track_db, ""); 
     }
+    return(track_db);
 }
 
 if (length(opt$hub_name) == 0) {
@@ -158,17 +162,22 @@
     hub_name <- opt$hub_name;
 }
 
-# Create the tracks output directory.
-tracks_output_dir = paste("tracks_", opt$hub_id, "/", sep="");
-dir.create(tracks_output_dir, showWarnings=FALSE);
-
 # Create the color scheme.
 state_color <- create_color_scheme(opt$input_dir_para);
 
 # Create the tracks.
-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);
+tracks_dir <- paste(opt$output_trackhub_files_path, "/tracks_", opt$hub_id, "/", sep="");
+dir.create(tracks_dir, showWarnings=FALSE);
+track_db <- create_track_db(opt$input_dir_state, opt$chrom_len_file, opt$header, tracks_dir, hub_id, hub_name, state_color, opt$state_name, opt$cell_info);
 
 # Write the outputs.
-write.table(track_db, opt$output_track_db, quote=F, row.names=F, col.names=F);
-write.table(c(paste("genome", opt$build), opt$output_build), paste(tracks_output_dir, "genomes_", opt$hub_id, ".txt", sep=""), quote=F, row.names=F, col.names=F);
-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$hub_id, ".txt", sep=""), quote=F, row.names=F, col.names=F);
+track_db_file_name <- paste(tracks_dir, "trackDb_", opt$hub_id, ".txt", sep="")
+write.table(track_db, file=track_db_file_name, quote=F, row.names=F, col.names=F);
+
+obj_name <- c(paste("genome", opt$build), paste("trackDb trackDb_", opt$hub_id, ".txt", sep=""))
+genomes_file_name <- paste(opt$output_hub_files_path, "genomes_", opt$hub_id, ".txt", sep="")
+write.table(obj_name, file=genomes_file_name, quote=F, row.names=F, col.names=F);
+
+obj_name <- c(paste("hub", opt$hub_id), paste("shortLabel", opt$hub_id), paste("longLabel", hub_name), paste("genomesFile genomes_", opt$hub_id, ".txt", sep=""), paste("email", opt$email))
+hub_file_name <- paste(opt$output_hub_files_path, "hub_", opt$hub_id, ".txt", sep="")
+write.table(obj_name, file=hub_file_name, quote=F, row.names=F, col.names=F);