R/kegg.R

In jasenfinch/composition: Composition Analyses for Metabolomics Data

#' @importFrom KEGGREST keggLink
#' @importFrom dplyr distinct
#' @importFrom tibble deframe
#' @importFrom stringr str_remove_all

keggCompounds <- function(organism = character()){
  
  if (length(organism) == 0) {
    compounds <- metabolites %>%
      getAccessions() %>%
      .$ID
  } else {
    enzymes <- keggLink(organism,'enzyme') %>%
      names()
    compounds <- keggLink('compound','enzyme') %>%
      {tibble(Enzyme = names(.),Compound = .)} %>%
      filter(Enzyme %in% enzymes) %>%
      select(Compound) %>%
      distinct() %>%
      deframe() %>%
      str_remove_all('cpd:')
  }
  return(compounds)
}

#' keggPIPs
#' @description Molecular formula putative ionisation products from KEGG.
#' @param MFs Molecular formulas and adducts to search. Should be a tibble containing two character columns named MF and Adduct.
#' @param organism KEGG organism ID
#' @param adductRules table containing adduct formation rules. Defaults to mzAnnotation::adducts().
#' @importFrom mzAnnotation filterACCESSIONS filterMF filterIP getAccessions adducts
#' @importFrom dplyr everything filter
#' @importFrom tibble is_tibble
#' @export

keggPIPs <- function(MFs, organism = character(), adductRules = adducts()){
  
  if (!is.data.frame(MFs)) {
    stop('MFs should be a tibble or data frame.')
  }
  
  if (!identical(colnames(MFs),c('MF','Adduct'))) {
    stop('MFs should contain the columns MF and Adduct with molecular formulas and their putative adducts respectively.')
  }
  
  compounds <- keggCompounds(organism)
  
  met <- metabolites %>%
    filterACCESSIONS(compounds)
  
  MFs %>%
    split(1:nrow(.)) %>%
    map(~{
      rule <- adductRules %>%
        filter(Name == .x$Adduct)
      met %>%
        filterMF(.x$MF) %>%
        filterIP(rule = rule$Rule) %>%
        getAccessions() %>%
        mutate(MF = .x$MF,Adduct = .x$Adduct) %>%
        select(ID,MF,Adduct,everything())
    }) %>%
    bind_rows()
}

#' keggPathways
#' @description Return the KEGG pathways for a give set of compound IDs.
#' @param IDs character vector of KEGG compound IDs
#' @param organism KEGG organism ID
#' @importFrom dplyr inner_join left_join
#' @importFrom stringr str_remove_all str_replace_all str_split_fixed str_c
#' @importFrom KEGGREST keggGet
#' @export
    

keggPathways <- function(IDs,organism = character()){
  pathways <- keggLink('compound','pathway') %>%
    {tibble(Pathway = names(.),Compound = str_remove_all(.,'cpd:'))} %>%
    filter(Compound %in% IDs)
  
  if (length(organism) > 0) {
    organismPathways <- keggLink(organism,'pathway') %>%
      {tibble(Pathway = names(.))} %>%
      distinct() %>%
      mutate(Pathway = str_replace_all(Pathway,organism,'map'))
    pathways <- pathways %>%
      inner_join(organismPathways, by = "Pathway")
  }
  
  pathwayInfo <- pathways$Pathway %>%
    unique() %>%
    split(ceiling(seq_along(.)/10)) %>%
    map(~{
      keggGet(.x) %>%
        map(~{
          if ('CLASS' %in% names(.x)) {
            cl <- .x$CLASS %>%
              {str_split_fixed(.,'; ',2)[1,]}
          } else {
            cl <- NA
          }
          
          tibble(Pathway = str_c('path:',.x$ENTRY),
                 Name = .x$NAME,
                 Process = cl[1],
                 Class = cl[2]
          )
        }) %>%
        bind_rows()
    }) %>%
    bind_rows()
  
  pathways %>%
    left_join(pathwayInfo, by = "Pathway") %>%
    mutate(Pathway = str_remove_all(Pathway,'path:')) %>%
    select(Compound,everything())
}