R/getActiveDSEA.R

In bigomics/SPACEconnect: Stratified Perturbation Analysis by Correlation Enrichment of connectivity

#' Create the drug set enrichment object that is used in the visualization
#'
#' @param mDrugEnrich a large matrix represents the gene expression fold change in the presence of different perturbations, either various drugs concentrations or other genetic engineering
#' @param DrugsAnnot is  matrix represents the drugs annotation that contains the targets, mechanism of action , clinical phase, disease area, and indication
#' @param mFC A matrix of differential gene expression fold-change of own experiment, the rows name of the matrix must be the genes names
#' @param contr is a character string represent the two compared conditions(contrast) as it is provided from the fold-change matrix
#' @return drug set enrichment object
#' @export
#'
#' @examples # from the data-sets provided as examples within the package load the .rda files
#' # DrugsAnnot, mDrugEnrich, mFC
#' dsea <- getActiveDSEA(mDrugEnrich, DrugsAnnot, mFC, contr = colnames(mFC)[1])
#'
getActiveDSEA <- function(mDrugEnrich, DrugsAnnot, mFC, contr = NULL) {
  rownames(DrugsAnnot) <- DrugsAnnot$pert_iname

  if (is.null(contr)) {
    contr <- colnames(mFC)[1]
  }

  dr <- computePeturbEnrichment(
    mFC = mFC, mDrugEnrich = mDrugEnrich, DrugsAnnot = DrugsAnnot,
    methods = "GSEA", nprune = 250, contrast = contr
  )
  drs <- dr[["GSEA"]]

  nes <- round(drs$X[, contr], 4)
  pv <- round(drs$P[, contr], 4)
  qv <- round(drs$Q[, contr], 4)
  drug <- rownames(drs$X)
  stats <- dr$stats
  annot <- dr$annot
  nes[is.na(nes)] <- 0
  qv[is.na(qv)] <- 1
  pv[is.na(pv)] <- 1


  ## compile results matrix
  jj <- match(toupper(drug), toupper(rownames(DrugsAnnot)))
  annot <- DrugsAnnot[jj, c("moa", "target")]
  dt <- data.frame(drug = drug, NES = nes, pval = pv, padj = qv, annot)

  sel <- which(dt$moa != "" | dt$target != "")
  dts <- dt[sel, , drop = FALSE]


  dsea <- list(table = dts, clust = dr$clust, stats = stats)

  return(dsea)
}