R/run_phobius.R

In EliLillyCo/surfaltr: Rapid Comparison of Surface Protein Isoform Membrane Topologies Through surfaltr

#' Create a data frame with the membrane locations of each amino acid in a
#' protein using Phobius
#'
#' This function creates a data frame with columns containing transcript
#' IDs and corresponding output from Phobius. The Phobius output includes a
#' location for each amino acid, with O representing extracellular, M
#' representing transmembrane, S representing signal, and i representing
#' intracellular.
#'
#' @usage run_phobius(AA_seq, fasta_file_name)
#' @param AA_seq A data frame outputted by the get_pairs function
#' containing the gene names, transcript IDs, APPRIS annotations, and
#' protein sequences for each transcript.
#' @param fasta_file_name Path to fasta file containing amino acid sequences
#' @return A data frame containing each transcript ID and the corresponding
#' membrane location for each amino acid in its sequence formatted as a string
#' @note In order for this function to work, there needs to be a .fasta file
#' containing the amino acid sequences for each transcript called "AA.fasta"
#' saved to the working directory. Additionally, the file saves a copy of the
#' returned data frame in csv format to the working directory.
#' @importFrom utils write.csv
#' @importFrom stringr str_split
#' @importFrom readr parse_number
#' @examples
#' tmhmm_folder_name <- "~/TMHMM2.0c"
#' if (check_tmhmm_install(tmhmm_folder_name)) {
#'     currwd <- getwd()
#'     AA_seq <- get_pairs(system.file("extdata", "crb1_example.csv",
#'         package = "surfaltr"
#'     ), TRUE, "mouse", TRUE)
#'     topo <- run_phobius(AA_seq, paste(getwd(), "/AA.fasta", sep = ""))
#'     setwd(currwd)
#' }
#' @export


run_phobius <- function(AA_seq, fasta_file_name) {
    phobius <- get_phobius(fasta_file_name)
    topo <- data.frame(
        "Transcript_ID" = character(nrow(phobius)),
        "Output" = character(nrow(phobius))
    )
    topo$Transcript_ID <- phobius$Name
    for (row in seq_len(nrow(phobius))) {
        regs <- as.data.frame(stringr::str_split(phobius[row, "prediction"], "/"))
        phobius[row, "Signal"] <- regs[1, ]
        phobius[row, "Surf_Protein"] <- regs[2, ]
        if (is.na(phobius[row, "Surf_Protein"])) {
            phobius[row, "Surf_Protein"] <- regs[1, ]
            phobius[row, "Signal"] <- NA
        }
        if (is.na(phobius[row, "Signal"]) == FALSE) {
            num_sig <- substr(
                phobius[row, "Signal"], nchar(phobius[row, "Signal"]) - 1,
                nchar(phobius[row, "Signal"])
            )
            phobius[row, "Mem_String"] <- paste(replicate(num_sig, "S"), collapse = "")
        }
        surf_prot <- as.data.frame(stringr::str_split(phobius[row, "Surf_Protein"], "-"))
        colnames(surf_prot) <- c("Split")
        for (val in seq_len(nrow(surf_prot))) {
            if ((substr(surf_prot[val, "Split"], 1, 1) == "i" | substr(surf_prot[val, "Split"], 1, 1) == "o") &&
                nchar(surf_prot[val, "Split"]) == 1) {
                curr_let <- substr(surf_prot[val, "Split"], 1, 1)
                num_lets <- nchar(AA_seq[row, "protein_sequence"])
                phobius[row, "Mem_String"] <- paste(replicate(num_lets, curr_let), collapse = "")
            }
            if ((substr(surf_prot[val, "Split"], 1, 1) == "i" | substr(surf_prot[val, "Split"], 1, 1) == "o") &&
                nchar(surf_prot[val, "Split"]) != 1) {
                curr_let <- substr(surf_prot[val, "Split"], 1, 1)
                num_lets <- readr::parse_number(surf_prot[val, "Split"])
                phobius[row, "Mem_String"] <- paste(replicate(num_lets, curr_let), collapse = "")
                nxt_len <- stringr::str_extract_all(surf_prot[val + 1, "Split"], "\\(?[0-9,.]+\\)?")[[1]]
                mem_len <- as.numeric(nxt_len[1]) - num_lets
                phobius[row, "Mem_String"] <- paste(phobius[row, "Mem_String"], paste(replicate(mem_len, "M"),
                    collapse = ""
                ), sep = "")
            }
            if (suppressWarnings(!is.na(as.numeric(substr(surf_prot[val, "Split"], 1, 1))))) {
                len_nums <- stringr::str_extract_all(surf_prot[val, "Split"], "\\(?[0-9,.]+\\)?")[[1]]
                if (length(len_nums) != 1) {
                    seg_len <- as.numeric(len_nums[2]) - as.numeric(len_nums[1])
                    curr_let <- gsub("[^a-zA-Z]", "", surf_prot[val, "Split"])
                    phobius[row, "Mem_String"] <- paste(phobius[row, "Mem_String"],
                        paste(replicate(seg_len, curr_let), collapse = ""),
                        sep = ""
                    )
                    nxt_len <- stringr::str_extract_all(surf_prot[val + 1, "Split"], "\\(?[0-9,.]+\\)?")[[1]]
                    mem_len <- as.numeric(nxt_len[1]) - as.numeric(len_nums[2])
                    phobius[row, "Mem_String"] <- paste(phobius[row, "Mem_String"],
                        paste(replicate(mem_len, "M"), collapse = ""),
                        sep = ""
                    )
                } else {
                    seg_len <- nchar(AA_seq[row, "protein_sequence"]) - as.numeric(len_nums)
                    curr_let <- gsub("[^a-zA-Z]", "", surf_prot[val, "Split"])
                    phobius[row, "Mem_String"] <- paste(phobius[row, "Mem_String"],
                        paste(replicate(seg_len, curr_let), collapse = ""),
                        sep = ""
                    )
                }
            }
        }
    }
    topo$Output <- phobius$Mem_String
    write.csv(topo, paste(getwd(), "/mem_topo_phobius.csv", sep = ""))
    return(topo)
}