Limma analysis: support of GPL files for getting gene annotations.

R/get_gene_annots_from_file.R

+#' @title Get gene annotations from a text file.
+#'
+#' @description This function reads a TSV file and extract the relevant
+#' information into a data-frame.
+#'
+#' So far, the key column (containing the probe_id) is always the first one and
+#' the other colums are all considered to contain relevant annotations. The
+#' first row is considered as the header row.
+#'
+#' @param folder A path to the folder where the file resides.
+#' @param filename The name of the file.
+#' @param entities The actual set of object ids for which we want gene annotations.
+#' @return A data-frame of the provided entities, annotated.
+get_gene_annots_from_file <- function(folder, filename, entities) {
+
+  # We define the full path.
+  file_path <- paste0(folder, filename)
+
+  # We read the file into a data-frame.
+  gene_annots <- read.delim(file_path, row.names = NULL, stringsAsFactors = FALSE)
+
+  # We do a bit of cleaning.
+  gene_annots[is.na(gene_annots)] <- ""
+  rownames(gene_annots)           <- gene_annots$PROBEID
+  gene_annots                     <- gene_annots[entities, ]
+  rownames(gene_annots)           <- seq(1, length(rownames(gene_annots)))
+
+  return(gene_annots)
+}

R/get_affy_annots.R → R/get_gene_annots_from_package.R

-#' @title Get Gene annotations for affy arrays.
+#' @title Get gene annotations from a BioConductor package.
 #'
 #' @description This function reads the library defined by the array type and
 #' extract the relevant information into a data-frame.
 #'
-#' @param array_type A string representing the array type.
+#' @param affy_library_name A string representing the library name (usually the array
+#' type followed by ".db".
 #' @param entities The actual set of object ids for which we want gene annotations.
 #' It can for instance be probe ids or gene ids. The parameter keytype has to be
 #' modified accordingly. By default, the expected type is probe id.
@@ -12,11 +13,10 @@
 #' @param keytype The type of key used. It will be used to map between the annotations
 #' and the provided list of entities. The default is probe id.
 #' @return A data-frame of the provided entities, annotated.
-get_affy_annots <- function(array_type, entities,
+get_gene_annots_from_package <- function(affy_library_name, entities,
                             columns = c("PROBEID", "SYMBOL", "ENSEMBL", "ENTREZID"),
                             keytype = "PROBEID") {
-  # We define the library name (removing underscores and adding .db).
-  affy_library_name <- paste0(stringr::str_to_lower(stringr::str_remove(array_type, "_")), ".db")
+  # We load the library given its name.
   cmd <- substitute(gene_annots <- AnnotationDbi::select(
     x       = local_db,
     keys    = entities,