R/create_hmmer_json_dataframe.R

In currocam/FascinRSCA: R and Python code for SCA with fascin protein

#' Converts a series of json files downloaded directly from HMMER into a DataFrame.
#'
#' @param hmmer_json_file A character vector with file paths.
#' @param file_label A character vector with a label with a label to identify the file they come from.
#' @param desc_null A Boolean, TRUE if you want to omit the 'desc' parameter which can cause problems.
#'
#' @return A DataFrame
#' @export
#'
#'

create_hmmer_json_dataframe <- function(hmmer_json_file, file_label,desc_null=TRUE) {
  ## Check arguments
  if(any(!file.exists(hmmer_json_file)))
    stop("Invalid file path/s")
  if(!is.character(file_label))
    stop("'file_label' should be character")
  if(length(hmmer_json_file) != length(file_label))
    stop("Both vectors should have the same length")
  purrr::map2_df(hmmer_json_file, file_label, function(x, y){
    x %>%
      readr::read_file() %>%
      rjson::fromJSON() %>%
      purrr::pluck("results") %>%
      purrr::pluck("hits") %>%
      purrr::map_dfr(~ .x %>%
                       purrr::when(desc_null ~ purrr::list_modify(.x, "desc" = NULL), .x) %>%
                       tibble::as_tibble()) %>%
      dplyr::mutate(file_label = y)
  })}