R/grin2.annotation.resources.R

In GRIN2: Genomic Random Interval (GRIN)

#' GRIN2 Annotation Resource Registry
#'
#' Internal registry defining the pre-generated annotation resources available
#' to GRIN2. For each resource, it stores the genome assembly, Ensembl release,
#' annotation bundle version, annotation type, file name, download URL, and
#' expected MD5 checksum.
#'
#' The registry is used by \code{get.ensembl.annotation()} and internal
#' annotation-retrieval functions to identify and verify the requested
#' annotation resource.
#'
#' @return A data frame describing the pre-generated annotation resources
#'   available to GRIN2.
#'
#' @keywords internal
#' @noRd
grin2.annotation.resources <- function()
{
  github.owner <- "abdel-elsayed87"
  github.repository <- "GRIN2-annotations"
  release.tag <- "grch38-ensembl110-v1.0.0"

  base.url <- paste0("https://github.com/", github.owner, "/",
                     github.repository, "/releases/download/",
                     release.tag, "/")

  file.names <- c("GRCh38_Ensembl110_gene_annotation.rds",
                  "GRCh38_Ensembl110_exon_annotation.rds",
                  "GRCh38_Ensembl110_regulatory_annotation.rds")

  data.frame(
    genome.assembly = rep("Human_GRCh38", 3L),
    ensembl.version = rep(110L, 3L),
    bundle.version = rep("1.0.0", 3L),
    annotation.type = c("gene", "exon", "regulatory"),
    file.name = file.names,
    url = paste0(base.url, file.names),
    md5 = c("c0299beedd54697a6084dcb0266e4f4d",
            "5e827b45220eaa98fc2370468d665074",
            "60dc4994758aa016898393419d3fb1bf"),
    stringsAsFactors = FALSE
  )
}