In vobencha/curatedMetagenomicData: Curated Metagenomic Data of the Human Microbiome

library(curatedMetagenomicData)

sampleMetadata |>
    dplyr::rename(Study = study_name) |>
    dplyr::rename(Curator = curator) |>
    dplyr::rename(Disease = disease) |>
    dplyr::rename(`Body Site` = body_site) |>
    dplyr::rename(`Body Subsite` = body_subsite) |>
    dplyr::select(Study, PMID, Curator, Disease, `Body Site`, `Body Subsite`) |>
    dplyr::mutate(PMID = stringr::str_replace(PMID, "unpublished", NA_character_)) |>
    dplyr::mutate(Curator = stringr::str_replace_all(Curator, ",", ";")) |>
    dplyr::group_by(Study) |>
    dplyr::summarize(
        Study = stringr::str_c(base::unique(Study), collapse = ";"),
        PMID = stringr::str_c(base::unique(PMID), collapse = ";"),
        Curator = stringr::str_c(base::unique(Curator), collapse = ";"),
        Disease = stringr::str_c(base::unique(Disease), collapse = ";"),
        `Body Site` = stringr::str_c(base::unique(`Body Site`), collapse = ";"),
        `Body Subsite` = stringr::str_c(base::unique(`Body Subsite`), collapse = ";"),
        Samples = dplyr::n()
    ) |>
    dplyr::group_by(Study) |>
    dplyr::mutate(
        PMID = stringr::str_c(base::unique(stringr::str_sort(stringr::str_split(PMID, ";", simplify = TRUE))), collapse = ", "),
        Curator = stringr::str_c(base::unique(stringr::str_split(Curator, ";", simplify = TRUE)), collapse = ", "),
        Disease = stringr::str_c(base::unique(stringr::str_sort(stringr::str_split(Disease, ";", simplify = TRUE))), collapse = ", "),
        `Body Site` = stringr::str_c(base::unique(stringr::str_sort(stringr::str_split(`Body Site`, ";", simplify = TRUE))), collapse = ", "),
        `Body Subsite` = stringr::str_c(base::unique(stringr::str_sort(stringr::str_split(`Body Subsite`, ";", simplify = TRUE))), collapse = ", ")
    ) |>
    dplyr::mutate(Curator = stringr::str_replace_all(Curator, "_", " ")) |>
    dplyr::mutate(Disease = stringr::str_replace_all(Disease, "_", " ")) |>
    dplyr::mutate(`Body Site` = stringr::str_replace_all(`Body Site`, "cavity", "_cavity")) |>
    dplyr::mutate(`Body Site` = stringr::str_replace_all(`Body Site`, "_", " ")) |>
    dplyr::mutate(`Body Subsite` = stringr::str_replace_all(`Body Subsite`, ", l_", ", left_")) |>
    dplyr::mutate(`Body Subsite` = stringr::str_replace_all(`Body Subsite`, ", r_", ", right_")) |>
    dplyr::mutate(`Body Subsite` = stringr::str_replace_all(`Body Subsite`, "_", " ")) |> 
    dplyr::ungroup() |>
    dplyr::mutate(dplyr::across(.cols = -Samples, .fns = ~ tidyr::replace_na(.x, ""))) |>
    DT::datatable(options = list(dom = "t", pageLength = 100), extensions = "Responsive")

vobencha/curatedMetagenomicData documentation built on Nov. 4, 2024, 7:43 a.m.