R/moa_conn.R

In girke-lab/signatureSearch: Environment for Gene Expression Searching Combined with Functional Enrichment Analysis

##' Summarize GESS Results on MOA Level
##' 
##' Function summarizes GESS results on Mode of Action (MOA) level.
##' It returns a tabular representation of MOA 
##' categories ranked by their average signature search similarity to a query
##' signature.
##' @param gess_tb tibble in \code{\link{gessResult}} object
##' @param moa_cats if set as "default", it uses MOA annotations from the CLUE 
##' website (https://clue.io). 
##' Users can customize it by providing a `list` of character vectors containing
##' drug names and MOA categories as list component names.
##' @param cells one of "normal", "cancer" or "all", or a character vector 
##' containing cell types of interest.
##' \itemize{
##'   \item "all": all cell types in LINCS database;
##'   \item "normal": normal cell types in LINCS database as one group; 
##'   \item "tumor": tumor cell types in LINCS database as one group;
##' }
##' @details 
##' Column description of the result table:
##' 
##' moa: Mechanism of Action (MOA)
##' 
##' cells: cell type information
##' 
##' mean_rank: mean rank of drugs in corresponding GESS result for each MOA 
##' category
##' 
##' n_drug: number of drugs in each MOA category
##' @return data.frame 
##' @seealso \code{\link{gessResult}}
##' @examples 
##' res_moa <- moa_conn(dplyr::tibble(
##'                     pert=c("vorinostat", "trichostatin-a", "HC-toxin"), 
##'                     cell=rep("SKB",3), 
##'                     pval=c(0.001,0.02,0.05)))
##' @export
moa_conn <- function(gess_tb, moa_cats="default", cells="normal"){
  if(moa_cats=="default"){
    data("clue_moa_list", envir = environment())
    moa_list <- clue_moa_list
  }
  if(is(moa_cats, "list")){
    moa_list <- moa_cats
  }
  moa_mrk <- moa_mrk(gess_tb, moa_list, cells=cells)
  return(moa_mrk)
}

##' @importFrom dplyr mutate
moa_mrk <- function(gess_tb, moa_list, cells){
  gess_tb %<>% mutate(rank=seq_len(nrow(gess_tb)))
  if(cells=="all"){
    moa_mrk <- sort(vapply(moa_list, function(x) 
      mean(as.numeric(gess_tb$rank[gess_tb$pert %in% x]), na.rm=TRUE),
      FUN.VALUE = numeric(1)))
    moa_mrk_df <- data.frame(moa=names(moa_mrk), cell="all", 
                             mean_rank=moa_mrk, row.names = NULL, 
                             stringsAsFactors = FALSE)
  }
  else if(cells=="normal"){
    cell_sel <- c("HA1E", "HCC515", "HEK293T", "MCF10A", "NKDBA", "NEU", "NPC", 
                  "FIBRNPC", "ASC", "CD34", "PHH", "SKB")
    gess_tb_sub <- filter(gess_tb, cell %in% cell_sel)
    moa_mrk <- sort(vapply(moa_list, function(x) 
      mean(as.numeric(gess_tb_sub$rank[gess_tb_sub$pert %in% x]), na.rm=TRUE),
      FUN.VALUE = numeric(1)))
    moa_mrk_df <- data.frame(moa=names(moa_mrk), cells="normal", 
                             mean_rank=moa_mrk, row.names = NULL, 
                             stringsAsFactors = FALSE)
  }
  else if(cells=="tumor"){
    cell_sel <- c("A375", "A549", "BT20", "HEPG2", "HL60", "HS578T", "HT29", 
                  "HUH7", "JURKAT", "MCF7", "MDAMB231", "NOMO1", "PC3",
                  "SKBR3", "THP1", "U266", "U937", "VCAP")
    gess_tb_sub <- filter(gess_tb, cell %in% cell_sel)
    moa_mrk <- sort(vapply(moa_list, function(x) 
      mean(as.numeric(gess_tb_sub$rank[gess_tb_sub$pert %in% x]), na.rm=TRUE),
      FUN.VALUE = numeric(1)))
    moa_mrk_df <- data.frame(moa=names(moa_mrk), cells="tumor", 
                             mean_rank=moa_mrk, row.names = NULL, 
                             stringsAsFactors = FALSE)
  } else {
    cell_sel <- cells
    gess_tb_sub <- filter(gess_tb, cell %in% cell_sel)
    moa_mrk <- sort(vapply(moa_list, function(x) 
      mean(as.numeric(gess_tb_sub$rank[gess_tb_sub$pert %in% x]), na.rm=TRUE),
      FUN.VALUE = numeric(1)))
    moa_mrk_df <- data.frame(moa=names(moa_mrk), cells="group", 
                             mean_rank=moa_mrk, row.names = NULL, 
                             stringsAsFactors = FALSE)
  }
  moa_num <- data.frame(moa=names(moa_list), 
                        n_drug=vapply(moa_list, length, FUN.VALUE = integer(1)),
                        stringsAsFactors = FALSE)
  res <- as_tibble(moa_mrk_df) %>% left_join(as_tibble(moa_num), by="moa")
  return(res)
}