biom2phyloseq.R

getMeta <- function(file, delim = NULL, col_types = NULL, original_sample_name = TRUE) {
    readr::read_delim(file = file, delim = delim, col_types = col_types, show_col_types = is.null(col_types)) |>
        dplyr::rename(Sample = 1) |>
        (\(df) if (original_sample_name) dplyr::mutate(df, original_sample_name = Sample) else df)() |>
        tibble::column_to_rownames("Sample")
}


getBiom <- function(biomfilename, biomformat = c("FROGS", "stdBIOM"), treefilename = NULL, refseqfilename = NULL, refseqFunction = Biostrings::readDNAStringSet, refseqArgs = NULL) {
    biomformat <- match.arg(biomformat)

    switch(biomformat,
        "FROGS" = phyloseq.extended::import_frogs(biom = biomfilename, treefilename = treefilename, refseqfilename = refseqfilename, refseqFunction = refseqFunction, refseqArgs = refseqArgs),
        "stdBIOM" = phyloseq::import_biom(BIOMfilename = biomfilename, treefilename = treefilename, refseqfilename = refseqfilename, refseqFunction = refseqFunction, refseqArgs = refseqArgs)
    )
}

rename_sample <- function(physeq, rename_file) {
    rename_df <- readr::read_csv(rename_file, col_types = readr::cols()) |> tibble::deframe()

    phyloseq::sample_names(physeq) <- phyloseq::sample_names(physeq) |>
        purrr::map_chr(~ rename_df[.x] %||% .x)
    physeq
}

biom2physeq <- function(
    biomfilename, biomformat = c("FROGS", "stdBIOM"),
    metafilename = NULL, delim = NULL, col_types = NULL, original_sample_name = TRUE,
    treefilename = NULL, refseqfilename = NULL, refseqFunction = Biostrings::readDNAStringSet, refseqArgs = NULL, rename_file = NULL) {
    physeq <- getBiom(biomfilename = biomfilename, biomformat = biomformat, treefilename = treefilename, refseqfilename = refseqfilename, refseqFunction = refseqFunction, refseqArgs = refseqArgs)

    if (!is.null(metafilename)) {
        phyloseq::sample_data(physeq) <- phyloseq::sample_data(getMeta(file = metafilename, delim = delim, col_types = col_types, original_sample_name = original_sample_name))
    }

    if (!is.null(rename_file)) {
        physeq <- rename_sample(physeq, rename_file)
    }
    return(physeq)
}