Encode Experiment ID -> Narrow Peak URLs

encExp2Peaks.R

encExp2Peaks <- function(acc, getIDR=FALSE) {
  resp <- httr::GET(url = paste0("https://www.encodeproject.org/search/?type=File&dataset=/experiments/",
                                 acc,"/&file_type=bed+narrowPeak&format=json&frame=object&limit=all"), httr::accept_json()) %>%
    httr::content(as="parsed")
  res <- lapply(seq(resp$`@graph`), function(i) {
    entry <- resp$`@graph`[[i]]
    tibble(
      accfile = entry$accession,
      sup = ! is.null(entry$superseded_by %>% unlist()),
      type = entry$file_type,
      assembly = entry$assembly,
      date = as.Date(entry$date_created),
      url = entry$s3_uri,
      output_type = entry$output_type,
      biol_reps = entry$biological_replicates %>% unlist() %>% paste0(collapse = "_"),
      acc = acc
    )
  }) %>% bind_rows() %>%
    mutate(best_samps = case_when(
      "1_2" %in% biol_reps ~ "1_2",
      TRUE ~ "1"
    )) %>%
    dplyr::filter(! sup,
                  assembly == "GRCh38",
                  biol_reps == best_samps) %>%
    mutate(dateMax = max(date)) %>%
    dplyr::filter(date == dateMax) %>%
    mutate(url = paste0("https://www.encodeproject.org/files/", accfile, "/@@download/", accfile, ".bed.gz"))
  if (nrow(res) > 1 & getIDR) {
    res <- dplyr::filter(res, output_type == "IDR thresholded peaks")
  } else {
    res
  }
}

Takes an encode experiment accession and pulls the latest GRCh38 narrow peak file URL (with 1_2 replicates combined, if available).

Usage:

 > encExp2Peaks("ENCSR825SVO")

# A tibble: 1 × 8
  accfile     sup   type           assembly url                                                                           biol_reps acc   best_samps
  <chr>       <lgl> <chr>          <chr>    <chr>                                                                         <chr>     <chr> <chr>     
1 ENCFF128BDG FALSE bed narrowPeak GRCh38   https://www.encodeproject.org/files/ENCFF128BDG/@@download/ENCFF128BDG.bed.gz 1_2       ENCS… 1_2