R/weight_bio.R

In taxonomicallyinformedannotation/tima-r: Taxonomically Informed Metabolite Annotation

#' @title Weight bio
#'
#' @description This function weights the eventually MS1
#' complemented annotations according their biological source
#'
#' @include transform_score_sirius_csi.R
#' @include log_pipe.R
#'
#' @param annotation_table_taxed Table containing the initial annotation
#' eventually complemented by additional MS1 annotations
#' @param structure_organism_pairs_table Table containing the
#' structure - organism pairs
#' @param weight_spectral Weight for the spectral score
#' @param weight_biological Weight for the biological score
#' @param score_biological_domain Score for a `domain` match
#' (should be lower than `kingdom`)
#' @param score_biological_kingdom Score for a `kingdom` match
#'  (should be lower than `phylum`)
#' @param score_biological_phylum Score for a `phylum` match
#'  (should be lower than `class`)
#' @param score_biological_class Score for a `class` match
#' (should be lower than `order`)
#' @param score_biological_order Score for a `order` match
#' (should be lower than `family`)
#' @param score_biological_family Score for a `family` match
#' (should be lower than `tribe`)
#' @param score_biological_tribe Score for a `tribe` match
#' (should be lower than `genus`)
#' @param score_biological_genus Score for a `genus` match
#' (should be lower than `species`)
#' @param score_biological_species Score for a `species` match
#'  (should be lower than `variety`)
#' @param score_biological_variety Score for a `variety` match
#' (should be the highest)
#'
#' @return A table containing the biologically weighted annotation
#'
#' @examples NULL
weight_bio <-
  function(annotation_table_taxed = get("annotation_table_taxed", envir = parent.frame()),
           structure_organism_pairs_table = get("structure_organism_pairs_table", envir = parent.frame()),
           weight_spectral = get("weight_spectral", envir = parent.frame()),
           weight_biological = get("weight_biological", envir = parent.frame()),
           score_biological_domain = get("score_biological_domain", envir = parent.frame()),
           score_biological_kingdom = get("score_biological_kingdom", envir = parent.frame()),
           score_biological_phylum = get("score_biological_phylum", envir = parent.frame()),
           score_biological_class = get("score_biological_class", envir = parent.frame()),
           score_biological_order = get("score_biological_order", envir = parent.frame()),
           score_biological_family = get("score_biological_family", envir = parent.frame()),
           score_biological_tribe = get("score_biological_tribe", envir = parent.frame()),
           score_biological_genus = get("score_biological_genus", envir = parent.frame()),
           score_biological_species = get("score_biological_species", envir = parent.frame()),
           score_biological_variety = get("score_biological_variety", envir = parent.frame())) {
    df0 <- annotation_table_taxed |>
      tidytable::distinct(
        candidate_structure_tax_cla_01kin,
        candidate_structure_tax_npc_01pat,
        candidate_structure_tax_cla_02sup,
        candidate_structure_tax_npc_02sup,
        candidate_structure_tax_cla_03cla,
        candidate_structure_tax_npc_03cla,
        candidate_structure_tax_cla_04dirpar
      ) |>
      tima:::log_pipe("adding \"notClassified\" \n") |>
      tidytable::mutate(tidytable::across(
        .cols = tidyselect::matches("candidate_structure_"),
        .fns = function(x) {
          tidytable::replace_na(x, "notClassified")
        }
      ))

    df1 <- annotation_table_taxed |>
      tidytable::select(
        candidate_structure_inchikey_no_stereo,
        sample_organism_01_domain,
        sample_organism_02_kingdom,
        sample_organism_03_phylum,
        sample_organism_04_class,
        sample_organism_05_order,
        # sample_organism_05_1_infraorder,
        sample_organism_06_family,
        # sample_organism_06_1_subfamily,
        sample_organism_07_tribe,
        # sample_organism_07_1_subtribe,
        sample_organism_08_genus,
        # sample_organism_08_1_subgenus,
        sample_organism_09_species,
        # sample_organism_09_1_subspecies,
        sample_organism_10_varietas
      ) |>
      tidytable::distinct() |>
      tidytable::left_join(
        structure_organism_pairs_table |>
          tidytable::filter(!is.na(structure_inchikey_no_stereo)) |>
          tidytable::select(
            candidate_structure_inchikey_no_stereo = structure_inchikey_no_stereo,
            candidate_organism_01_domain = organism_taxonomy_01domain,
            candidate_organism_02_kingdom = organism_taxonomy_02kingdom,
            candidate_organism_03_phylum = organism_taxonomy_03phylum,
            candidate_organism_04_class = organism_taxonomy_04class,
            candidate_organism_05_order = organism_taxonomy_05order,
            # candidate_organism_05_1_infraorder =
            # organism_taxonomy_05_1infraorder,
            candidate_organism_06_family = organism_taxonomy_06family,
            # candidate_organism_06_1_subfamily =
            # organism_taxonomy_06_1subfamily,
            candidate_organism_07_tribe = organism_taxonomy_07tribe,
            # candidate_organism_07_1_subtribe = organism_taxonomy_07_1subtribe,
            candidate_organism_08_genus = organism_taxonomy_08genus,
            # candidate_organism_08_1_subgenus = organism_taxonomy_08_1subgenus,
            candidate_organism_09_species = organism_taxonomy_09species,
            # candidate_organism_09_1_subspecies =
            # organism_taxonomy_09_1subspecies,
            candidate_organism_10_varietas = organism_taxonomy_10varietas
          ) |>
          tidytable::distinct() |>
          tidytable::mutate(tidytable::across(
            .cols = tidyselect::where(is.character),
            .fns = function(x) {
              tidytable::na_if(x, "")
            }
          ))
      )

    df2 <- df1 |>
      tidytable::distinct(
        sample_organism_01_domain,
        sample_organism_02_kingdom,
        sample_organism_03_phylum,
        sample_organism_04_class,
        sample_organism_05_order,
        # sample_organism_05_1_infraorder,
        sample_organism_06_family,
        # sample_organism_06_1_subfamily,
        sample_organism_07_tribe,
        # sample_organism_07_1_subtribe,
        sample_organism_08_genus,
        # sample_organism_08_1_subgenus,
        sample_organism_09_species,
        # sample_organism_09_1_subspecies,
        sample_organism_10_varietas,
        candidate_organism_01_domain,
        candidate_organism_02_kingdom,
        candidate_organism_03_phylum,
        candidate_organism_04_class,
        candidate_organism_05_order,
        # candidate_organism_05_1_infraorder,
        candidate_organism_06_family,
        # candidate_organism_06_1_subfamily,
        candidate_organism_07_tribe,
        # candidate_organism_07_1_subtribe,
        candidate_organism_08_genus,
        # candidate_organism_08_1_subgenus,
        candidate_organism_09_species,
        # candidate_organism_09_1_subspecies,
        candidate_organism_10_varietas
      )

    log_debug("calculating biological score at all levels ... \n")
    score_per_level_bio <-
      function(df,
               candidates,
               samples,
               score,
               score_name) {
        score <- df |>
          tidytable::distinct(!!as.name(candidates), !!as.name(samples)) |>
          tidytable::filter(!is.na(!!as.name(samples))) |>
          tidytable::filter(!!as.name(samples) != "ND") |>
          tidytable::filter(!is.na(!!as.name(candidates))) |>
          tidytable::filter(!!as.name(candidates) != "notClassified") |>
          tidytable::filter(
            stringi::stri_detect_regex(
              pattern = !!as.name(candidates),
              str = !!as.name(samples)
            ) |
              !!as.name(samples) == "notClassified"
          ) |>
          tidytable::mutate(
            !!as.name(score_name) := tidytable::if_else(
              condition = !!as.name(samples) != "notClassified",
              true = !!as.name(score) * 1,
              false = 0
            )
          )
      }

    log_debug("... domain \n")
    step_dom <- df2 |>
      score_per_level_bio(
        candidates = "candidate_organism_01_domain",
        samples = "sample_organism_01_domain",
        score = "score_biological_domain",
        score_name = "score_biological_01"
      )
    log_debug("... kingdom \n")
    step_kin <- df2 |>
      score_per_level_bio(
        candidates = "candidate_organism_02_kingdom",
        samples = "sample_organism_02_kingdom",
        score = "score_biological_kingdom",
        score_name = "score_biological_02"
      )
    log_debug("... phylum \n")
    step_phy <- df2 |>
      score_per_level_bio(
        candidates = "candidate_organism_03_phylum",
        samples = "sample_organism_03_phylum",
        score = "score_biological_phylum",
        score_name = "score_biological_03"
      )
    log_debug("... class \n")
    step_cla <- df2 |>
      score_per_level_bio(
        candidates = "candidate_organism_04_class",
        samples = "sample_organism_04_class",
        score = "score_biological_class",
        score_name = "score_biological_04"
      )
    log_debug("... order \n")
    step_ord <- df2 |>
      score_per_level_bio(
        candidates = "candidate_organism_05_order",
        samples = "sample_organism_05_order",
        score = "score_biological_order",
        score_name = "score_biological_05"
      )
    # log_debug("... infraorder \n")
    # step_ord2 <- df2 |>
    #   score_per_level_bio(
    #     candidates = "candidate_organism_05_1_infraorder",
    #     samples = "sample_organism_05_1_infraorder",
    #     score = "score_biological_infraorder",
    #     score_name = "score_biological_05_1"
    #   )
    log_debug("... family \n")
    step_fam <- df2 |>
      score_per_level_bio(
        candidates = "candidate_organism_06_family",
        samples = "sample_organism_06_family",
        score = "score_biological_family",
        score_name = "score_biological_06"
      )
    # log_debug("... subfamily \n")
    # step_fam2 <- df2 |>
    #   score_per_level_bio(
    #     candidates = "candidate_organism_06_1_subfamily",
    #     samples = "sample_organism_06_1_subfamily",
    #     score = "score_biological_subfamily",
    #     score_name = "score_biological_06_1"
    #   )
    log_debug("... tribe \n")
    step_tri <- df2 |>
      score_per_level_bio(
        candidates = "candidate_organism_07_tribe",
        samples = "sample_organism_07_tribe",
        score = "score_biological_tribe",
        score_name = "score_biological_07"
      )
    # log_debug("... subtribe \n")
    # step_tri2 <- df2 |>
    #   score_per_level_bio(
    #     candidates = "candidate_organism_07_1_subtribe",
    #     samples = "sample_organism_07_1_subtribe",
    #     score = "score_biological_subtribe",
    #     score_name = "score_biological_07_1"
    #   )
    log_debug("... genus \n")
    step_gen <- df2 |>
      score_per_level_bio(
        candidates = "candidate_organism_08_genus",
        samples = "sample_organism_08_genus",
        score = "score_biological_genus",
        score_name = "score_biological_08"
      )
    # log_debug("... subgenus \n")
    # step_gen2 <- df2 |>
    #   score_per_level_bio(
    #     candidates = "candidate_organism_08_1_subgenus",
    #     samples = "sample_organism_08_1_subgenus",
    #     score = "score_biological_subgenus",
    #     score_name = "score_biological_08_1"
    #   )
    log_debug("... species \n")
    step_spe <- df2 |>
      score_per_level_bio(
        candidates = "candidate_organism_09_species",
        samples = "sample_organism_09_species",
        score = "score_biological_species",
        score_name = "score_biological_09"
      )
    # log_debug("... subspecies \n")
    # step_spe2 <- df2 |>
    #   score_per_level_bio(
    #     candidates = "candidate_organism_09_1_subspecies",
    #     samples = "sample_organism_09_1_subspecies",
    #     score = "score_biological_subspecies",
    #     score_name = "score_biological_09_1"
    #   )
    log_debug("... varietas \n")
    step_var <- df2 |>
      score_per_level_bio(
        candidates = "candidate_organism_10_varietas",
        samples = "sample_organism_10_varietas",
        score = "score_biological_variety",
        score_name = "score_biological_10"
      )

    log_debug("... keeping best biological score \n")
    annot_table_wei_bio <- df2 |>
      tidytable::left_join(step_dom) |>
      tidytable::left_join(step_kin) |>
      tidytable::left_join(step_phy) |>
      tidytable::left_join(step_cla) |>
      tidytable::left_join(step_ord) |>
      # tidytable::left_join(step_ord2) |>
      tidytable::left_join(step_fam) |>
      # tidytable::left_join(step_fam2) |>
      tidytable::left_join(step_tri) |>
      # tidytable::left_join(step_tri2) |>
      tidytable::left_join(step_gen) |>
      # tidytable::left_join(step_gen2) |>
      tidytable::left_join(step_spe) |>
      # tidytable::left_join(step_spe2) |>
      tidytable::left_join(step_var) |>
      tidytable::mutate(
        score_biological = pmax(
          score_biological_01,
          score_biological_02,
          score_biological_03,
          score_biological_04,
          score_biological_05,
          # score_biological_05_1,
          score_biological_06,
          # score_biological_06_1,
          score_biological_07,
          # score_biological_07_1,
          score_biological_08,
          # score_biological_08_1,
          score_biological_09,
          # score_biological_09_1,
          score_biological_10,
          0,
          na.rm = TRUE
        )
      ) |>
      tidytable::select(-tidyselect::contains("score_biological_")) |>
      tidytable::mutate(
        candidate_structure_organism_occurrence_closest = tidytable::case_when(
          score_biological == score_biological_domain
          ~ candidate_organism_01_domain,
          score_biological == score_biological_kingdom
          ~ candidate_organism_02_kingdom,
          score_biological == score_biological_phylum
          ~ candidate_organism_03_phylum,
          score_biological == score_biological_class
          ~ candidate_organism_04_class,
          score_biological == score_biological_order
          ~ candidate_organism_05_order,
          # score_biological == score_biological_05_1 ~
          # candidate_organism_05_1_infraorder,
          score_biological == score_biological_family
          ~ candidate_organism_06_family,
          # score_biological == score_biological_06_1 ~
          # candidate_organism_06_1_subfamily,
          score_biological == score_biological_tribe
          ~ candidate_organism_07_tribe,
          # score_biological == score_biological_07_1 ~
          # candidate_organism_07_1_subtribe,
          score_biological == score_biological_genus
          ~ candidate_organism_08_genus,
          # score_biological == score_biological_08_1 ~
          # candidate_organism_08_1_subgenus,
          score_biological == score_biological_species
          ~ candidate_organism_09_species,
          # score_biological == score_biological_09_1 ~
          # candidate_organism_09_1_subspecies,
          score_biological == score_biological_variety ~
            candidate_organism_10_varietas,
          .default = NA_character_
        )
      ) |>
      tidytable::right_join(df1) |>
      tidytable::arrange(tidytable::desc(score_biological)) |>
      tidytable::distinct(
        candidate_structure_inchikey_no_stereo,
        sample_organism_01_domain,
        sample_organism_02_kingdom,
        sample_organism_03_phylum,
        sample_organism_04_class,
        sample_organism_05_order,
        # sample_organism_05_1_infraorder,
        sample_organism_06_family,
        # sample_organism_06_1_subfamily,
        sample_organism_07_tribe,
        # sample_organism_07_1_subtribe,
        sample_organism_08_genus,
        # sample_organism_08_1_subgenus,
        sample_organism_09_species,
        # sample_organism_09_1_subspecies,
        sample_organism_10_varietas,
        .keep_all = TRUE
      ) |>
      tidytable::inner_join(annotation_table_taxed) |>
      tidytable::select(
        -tidyselect::contains("candidate_organism"),
        -tidyselect::contains("sample_organism")
      ) |>
      tidytable::left_join(df0) |>
      tima:::log_pipe("... calculating weighted biological score \n") |>
      tidytable::mutate(candidate_score_sirius_csi_tmp = tima:::transform_score_sirius_csi(candidate_score_sirius_csi)) |>
      tidytable::mutate(
        candidate_score_pseudo_initial = tidytable::case_when(
          !is.na(candidate_score_similarity) &
            !is.na(candidate_score_sirius_csi) ~ (as.numeric(candidate_score_similarity) + candidate_score_sirius_csi_tmp) / 2,
          !is.na(candidate_score_similarity) ~ as.numeric(candidate_score_similarity),
          !is.na(candidate_score_sirius_csi) ~ candidate_score_sirius_csi_tmp,
          TRUE ~ 0
        )
      ) |>
      tidytable::select(-candidate_score_sirius_csi_tmp) |>
      tidytable::mutate(
        score_weighted_bio = (1 / (weight_biological + weight_spectral)) * weight_biological * score_biological + (1 / (weight_biological + weight_spectral)) * weight_spectral * candidate_score_pseudo_initial
      )

    rm(
      annotation_table_taxed,
      df0,
      df1,
      df2,
      step_dom,
      step_kin,
      step_phy,
      step_cla,
      step_ord,
      ## step_ord2,
      step_fam,
      ## step_fam2,
      step_tri,
      ## step_tri2,
      step_gen,
      ## step_gen2,
      step_spe,
      ## step_spe2,
      step_var
    )

    return(annot_table_wei_bio)
  }