R/read_magma_genes_raw.R

In NathanSkene/MAGMA_Celltyping: Find Causal Cell-Types Underlying Complex Trait Genetics

#' Read a \emph{.genes.raw} file 
#' 
#' Read a \emph{.genes.raw} file  generated by
#'  \link[MAGMA.Celltyping]{map_snps_to_genes} 
#'  as a \link[data.table]{data.table}.
#' @param path Path to a MAGMA \emph{.genes.raw} file.
#' @param verbose Print messages.
#' @returns \link[data.table]{data.table}
#' 
#' @keywords internal
#' @importFrom stringr str_split
#' @importFrom data.table data.table rbindlist setnames :=
#' @importFrom dplyr relocate
read_magma_genes_raw <- function(path,
                                 verbose = TRUE){
    messager("Importing genes.raw file.",v=verbose)
    entrez <- hgnc_symbol <- NULL;
    l <- readLines(path)[-seq_len(2)] 
    l <- stringr::str_split(l," ")
    dt <- data.table::rbindlist(
        lapply(l, function(x) data.table::data.table(t(x))),
        fill = TRUE
    )
    data.table::setnames(dt,
                         old = c("V1","V2","V3","V4"),
                         new = c("entrez","chrom","start","end")) 
    dt[,entrez:=as.character(entrez)]
    dt <- merge(
        x = dt,
        y = MAGMA.Celltyping::hgnc2entrez_orthogene,
        by = "entrez"
    )
    dt <- dplyr::relocate(dt,hgnc_symbol, .after = "entrez")
    return(dt)
}