Mercurial > repos > kyost > atac_primer_tool

annotate make_window_bed.R @ 4:72571a30f17b draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add test for unique peak names.
author kyost
date Wed, 02 May 2018 13:20:21 -0400
parents fd3ea97a96bc
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
1 ## Command to run tool:
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
2 # Rscript --vanilla make_window_bed.R qPCR_peaks.bed window_size output_file
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
3
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
4 # Set up R error handling to go to stderr
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
5 options(show.error.messages=F, error=function(){cat(geterrmessage(),file=stderr());q("no",1,F)})
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
6
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
7 # Avoid crashing Galaxy with an UTF8 error on German LC settings
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
8 loc <- Sys.setlocale("LC_MESSAGES", "en_US.UTF-8")
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
9
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
10 args <- commandArgs(TRUE)
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
11
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
12 qPCR_bed <- args[1]
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
13 window_size <- strtoi(args[2])
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
14 output_file <- args[3]
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
15
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
16 qPCR_table <- read.delim(qPCR_bed, header=FALSE, stringsAsFactors=FALSE)
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
17
4
72571a30f17b Add test for unique peak names.
kyost
parents: 0
diff changeset
18 qPCR_table[,4] <- make.unique(qPCR_table[,4], sep = "_")
72571a30f17b Add test for unique peak names.
kyost
parents: 0
diff changeset
19
0
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
20 make_windows_bed <- function(a, size) { #a is a bed file containing peaks of interest and coordinates
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
21 #generates bed file with overlapping windows spanning each peak of specified size
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
22 return_bed <- data.frame()
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
23 for (i in 1:nrow(a)) {
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
24 line <- data.frame(a[i,],stringsAsFactors=FALSE)
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
25 peak_name <- as.character(line[1,4])
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
26 line <- rbind(line
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
27 , data.frame(V1=as.character(line[1,1])
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
28 , V2=line[1,2]
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
29 , V3=line[1,2]+size
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
30 , V4=as.character(paste(as.character(line[1,4])
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
31 , "_window1"
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
32 , sep = "")
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
33 )
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
34 )
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
35 )
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
36 count <- as.numeric(line[1,2])+size
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
37 nline <- 2
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
38 while (count < as.numeric(line[1,3])) {
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
39 line <- rbind(line
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
40 , data.frame(V1=as.character(line[nline,1])
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
41 , V2=line[nline,2]+floor(size/4)
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
42 , V3=line[nline,2]+floor(size/4)+size
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
43 , V4=as.character(paste(as.character(line[1,4])
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
44 , "_window"
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
45 , as.character(nline)
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
46 , sep = "")
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
47 )
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
48 )
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
49 )
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
50 nline <- nline + 1
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
51 count <- as.numeric(line[nline,3])
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
52 }
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
53 return_bed <- rbind(return_bed,line)
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
54 }
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
55 return(return_bed)
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
56 }
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
57
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
58 output <- make_windows_bed(qPCR_table, window_size)
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
59
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
60
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
61 write.table(output
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
62 , output_file
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
63 , sep = "\t"
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
64 , col.names = FALSE
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
65 , row.names = FALSE
fd3ea97a96bc planemo upload commit 103cb51ec368438642504c3f98b76c363db478bb
kyost
parents:
diff changeset
66 , quote = FALSE)