Download the latest GENCODE annotations file and export as a GRanges object

gencode-granges.r

# Download the latest GENCODE annotations file and export as a GRanges object
library(RCurl)
library(rtracklayer)


# Variables ---------------------------------------------------------------

# 'human' or 'mouse'
organism <- "human"

# output directory
out.dir <- "~/Downloads"


# Functions ---------------------------------------------------------------

ls_url <- function(url) {
  stopifnot(url.exists(url))
  out <- getURL(url, ftp.use.epsv = FALSE, dirlistonly = TRUE)
  readLines(textConnection(out))
}


# Determine latest release ------------------------------------------------

ftp.url <- paste0("ftp://ftp.sanger.ac.uk/pub/gencode/Gencode_", organism)
ftp.dirs <- ls_url(ftp.url)

ftp.dirs <- Filter(function(x) grepl("release", x), ftp.dirs)
releases <-sapply(strsplit(ftp.dirs, "_"), function(x) x[2])

ftp.dirs <- ftp.dirs[order(as.numeric(sub("\\D+", "", releases)), 
                           sub("\\d+", "", releases))]

message("Latest release is ", tail(ftp.dirs, 1))


# Download latest annotation.gtf file -------------------------------------

ftp.url <- paste0(ftp.url, "/", tail(ftp.dirs, 1), "/")
ftp.files <- ls_url(ftp.url)

gtf.file <- Filter(function(x) grepl("\\d\\.annotation.gtf", x), ftp.files)
gtf.url <- paste0(ftp.url, gtf.file)

download.file(gtf.url, destfile = file.path(out.dir, gtf.file))


# Export GRanges object ---------------------------------------------------

gencode <- import.gff(file.path(out.dir, gtf.file), version = "2")

save(gencode, file = file.path(out.dir, paste0(sub("gtf.gz", "", gtf.file), "rda")))