|
a |
|
b/R/downloadAndPrepareCELFiles.R |
|
|
1 |
#' Download GEO .CEL files |
|
|
2 |
#' |
|
|
3 |
#' Function for downloading the CEL files of a GEO dataset and prepare it for |
|
|
4 |
#' preprocessing. |
|
|
5 |
#' The function uses \code{getGEOSuppFiles} which downloads the data. The |
|
|
6 |
#' function then unzips the data files in the GEO series. |
|
|
7 |
#' |
|
|
8 |
#' @param geo_nbr The GEO ascession number. |
|
|
9 |
#' @param destdir The destination dir of the downloaded files. |
|
|
10 |
#' @param clean Should the strictly unnessesary files be deleted? |
|
|
11 |
#' @param verbose Signal the process. |
|
|
12 |
#' @note The function will overwrite existing files in the \code{destdir}. |
|
|
13 |
#' @author |
|
|
14 |
#' Anders Ellern Bilgrau, |
|
|
15 |
#' Steffen Falgreen Larsen |
|
|
16 |
#' @examples |
|
|
17 |
#' \dontrun{ |
|
|
18 |
#' downloadAndPrepareCELFiles(geo_nbr = "GSE18376", |
|
|
19 |
#' destdir = tempdir()) |
|
|
20 |
#' } |
|
|
21 |
#' @importFrom GEOquery getGEOSuppFiles gunzip |
|
|
22 |
#' @keywords internal |
|
|
23 |
#' @export |
|
|
24 |
downloadAndPrepareCELFiles <- function(geo_nbr, |
|
|
25 |
destdir = getwd(), |
|
|
26 |
clean = FALSE, |
|
|
27 |
verbose = TRUE) { |
|
|
28 |
options("timeout" = 240) # Increase timeout (in seconds) |
|
|
29 |
on.exit(options("timeout" = 60)) # Reset to standard timeout on exit |
|
|
30 |
|
|
|
31 |
if (verbose) message("Preparing GEO ", geo_nbr, " data") |
|
|
32 |
dl_dir <- file.path(destdir, geo_nbr) |
|
|
33 |
|
|
|
34 |
# Download data if nessesary |
|
|
35 |
raw_file <- paste0(file.path(dl_dir, geo_nbr), "_RAW.tar") |
|
|
36 |
if (!file.exists(raw_file)) { |
|
|
37 |
if (verbose) message("Downloading .CEL files...\n") |
|
|
38 |
dir.create(dl_dir, showWarnings = FALSE) |
|
|
39 |
getGEOSuppFiles(GEO = geo_nbr, makeDirectory = FALSE, baseDir = dl_dir) |
|
|
40 |
} else { |
|
|
41 |
if (verbose) message("Compressed .CEL files already downloaded...") |
|
|
42 |
} |
|
|
43 |
|
|
|
44 |
# List celfiles (if any) |
|
|
45 |
cel_files <- list.files(dl_dir, pattern = "\\.cel$", |
|
|
46 |
ignore.case = TRUE, full.names = TRUE) |
|
|
47 |
|
|
|
48 |
if (identical(cel_files, character(0))) { |
|
|
49 |
|
|
|
50 |
# Untar the file bundle |
|
|
51 |
if (verbose) message("Untaring the RAW file...\n") |
|
|
52 |
tar_file <- list.files(path = dl_dir, pattern = "RAW.tar$", full.names = TRUE) |
|
|
53 |
untar(tarfile = tar_file, exdir = dl_dir) |
|
|
54 |
|
|
|
55 |
# Unzip the files |
|
|
56 |
if (verbose) message("Unzipping the .CEL files...\n") |
|
|
57 |
gz_files <- list.files(dl_dir, pattern = "\\.cel.gz$", ignore.case = TRUE, |
|
|
58 |
full.names = TRUE) |
|
|
59 |
for (file in gz_files) { |
|
|
60 |
gunzip(file, overwrite = TRUE, remove = TRUE) |
|
|
61 |
} |
|
|
62 |
|
|
|
63 |
# List celfiles |
|
|
64 |
cel_files <- list.files(dl_dir, pattern = "\\.cel$", |
|
|
65 |
ignore.case = TRUE, full.names = TRUE) |
|
|
66 |
} else { |
|
|
67 |
message("(Some) .CEL files already unpacked...\n") |
|
|
68 |
} |
|
|
69 |
|
|
|
70 |
# Clean-up if wanted |
|
|
71 |
if (clean) { |
|
|
72 |
if (verbose) message("Removing all non CEL files...\n") |
|
|
73 |
file.remove(setdiff(list.files(dl_dir, full.names = TRUE), cel_files)) |
|
|
74 |
} |
|
|
75 |
|
|
|
76 |
cel_files <- normalizePath(cel_files) |
|
|
77 |
if (verbose) message("Download and preparation of CEL files done.\n") |
|
|
78 |
return(invisible(cel_files)) |
|
|
79 |
} |