ToxicoGx: Analysis of Large-Scale Toxico-Genomic Data

# tSet molecularProfiles from eSets to SEs
#
# Converts all ExpressionSet objects within the molecularProfiles slot of a
#   ToxicoSet to SummarizedExperiments
#
# @param tSet \code{S4} A ToxicoSet containing molecular data in ExpressionSets
#
# @return \code{S4} A ToxicoSet containing molecular data in a SummarizedExperiments
#
#' @importFrom SummarizedExperiment assay assays assayNames
#' @importClassesFrom SummarizedExperiment SummarizedExperiment Assays
#' @importFrom Biobase exprs fData pData annotation protocolData assayData experimentData
#' @importFrom S4Vectors SimpleList DataFrame
#' @importFrom stats setNames
#' @export
#' @keywords internal
#' @noRd
.convertTsetMolecularProfilesToSE <- function(tSet) {

  eSets <- molecularProfilesSlot(tSet) # Extract eSet data

  molecularProfilesSlot(tSet) <-
    lapply(eSets,
           function(eSet){

             # Build summarized experiment from eSet
             SE <- SummarizedExperiment::SummarizedExperiment(
               ## TODO:: Do we want to pass an environment for better memory efficiency?
               assays=SimpleList(as.list(Biobase::assayData(eSet))
               ),
               # Switch rearrange columns so that IDs are first, probes second
               rowData=S4Vectors::DataFrame(Biobase::fData(eSet),
                                            rownames=rownames(Biobase::fData(eSet))
               ),
               colData=S4Vectors::DataFrame(Biobase::pData(eSet),
                                            rownames=rownames(Biobase::pData(eSet))
               ),
               metadata=list("experimentData" = eSet@experimentData,
                             "annotation" = Biobase::annotation(eSet),
                             "protocolData" = Biobase::protocolData(eSet)
               )
             )
             ## TODO:: Determine if this can be done in the SE constructor?
             # Extract names from expression set
             SummarizedExperiment::assayNames(SE) <- Biobase::assayDataElementNames(eSet)
             # Assign SE to tSet
             mDataType <- Biobase::annotation(eSet)
             molecularProfilesSlot(tSet)[[mDataType]] <- SE
           })
  setNames(molecularProfilesSlot(tSet), names(eSets))
  tSet
}

# Validate tSet molecularProfiles Conversion
#
# Checks that all the information contained in an ExpressionSet molecularProfile
#   was successfully tranferred to the SummarizedExperiment molecularProfile
#
# @param tSet_new \code{S4} a tSet containing molecularProfiles as SummarizedExperiments
# @param tSet_old \code{S4} a tSet containing molecularProfiles as ExpressionSets
#
# @return \code{message} Any slots which are not the same
#
##' @importFrom assertthat are_equal
##' @importFrom SummarizedExperiment SummarizedExperiment Assays assay
##'   assayNames assayNames<-
##' @importFrom Biobase exprs fData pData annotation protocolData
##'   assayDataElementNames experimentData assayData
##' @keywords internal
#.validateTsetMolecularProfilesToSEConversion <- function(tSet_old, tSet_new) {
#
#  # Testing that tSets are in correct order
#  message("Checking is tSet structures are correct")
#
#  if(!all(vapply(tSet_old@molecularProfiles,
#                 function(x) { is(x, "ExpressionSet") },
#                 FUN.VALUE = logical(1)))
#  ) message("Old tSet doesn't contain ExpressionSet objects, maybe argument
#            order is wrong?")
#
#  if(!all(vapply(molecularProfilesSlot(tSet_new),
#                 function(x) { is(x, "SummarizedExperiment") },
#                 FUN.VALUE = logical(1)))
#  ) message("New tSet doesn't contain SummarizedExperiment objects, maybe
#            argument order is wrong?")
#
#  # Comparing molecularProfiles slot data
#  message("Checking molecularProfiles slots hold equivalent data.")
#
#  for (i in seq_len(length(tSet_old@molecularProfiles))) {
#    for (j in seq_along(assays(molecularProfilesSlot(tSet_new)[[i]]))) {
#      if(!all(as.list(assayData(tSet_old@molecularProfiles[[i]]))[[j]] ==
#              as.list(assays(molecularProfilesSlot(tSet_new)[[i]]))[[j]],
#          na.rm = TRUE)
#        ) message("The assay data is not equivalent")
#    }
#  }
#  ## TODO:: Rewrite this as an apply statement
#  for (i in seq_len(length(tSet_old@molecularProfiles))) { # Have to compare like this due to NAs in data
#    # Checking phenoData
#    if(
#      if (nrow(pData(tSet_old@molecularProfiles[[i]])) > 0) {
#        !all(
#          as(tSet_old@molecularProfiles[[i]]@phenoData, "data.frame") ==
#            as.data.frame(molecularProfilesSlot(tSet_new)[[i]]@colData[
#              seq_len(length(molecularProfilesSlot(tSet_new)[[i]]@colData) - 1)]),
#          na.rm = TRUE)
#      } else { FALSE }
#    ) message("The phenoData is not equivalent")
#    # Checking featureData
#    if(
#      if (nrow(fData(tSet_old@molecularProfiles[[i]])) > 0) {
#        !all(
#          as(tSet_old@molecularProfiles[[i]]@featureData, "data.frame") ==
#            as.data.frame(molecularProfilesSlot(tSet_new)[[i]]@elementMetadata[
#              seq_len(length(molecularProfilesSlot(tSet_new)[[i]]@elementMetadata) - 1)]),
#          na.rm=TRUE)
#      } else { FALSE }
#    ) message("The featureData is not equivalent")
#    # Checking protocolData
#    if(
#      !all(
#        as(tSet_old@molecularProfiles[[i]]@protocolData, "data.frame") ==
#          as(molecularProfilesSlot(tSet_new)[[i]]@metadata$protocolData, "data.frame"),
#        na.rm = TRUE)
#    ) message("The protocolData is not equivalent")
#  }
#
#  if(!assertthat::are_equal(
#    lapply(tSet_old@molecularProfiles, function(x) { annotation(x) }),
#    lapply(molecularProfilesSlot(tSet_new), function(x) { metadata(x)$annotation }))
#  ) message("The annotation is not equivalent")
#
#  if(!assertthat::are_equal(
#    lapply(tSet_old@molecularProfiles, function(x) { experimentData(x) }),
#    lapply(molecularProfilesSlot(tSet_new), function(x) { metadata(x)$experimentData }))
#  ) message("The experimentData is not equivalent")
#
#  # Comparing remainder of tSet slots; should not be affect by conversion
#  message("Comparing remainder of tSet slots")
#
#  if(!assertthat::are_equal(annotation(tSet_old), annotation(tSet_new)))
#    message("The annotation slots are not equivalent!")
#
#  if(!assertthat::are_equal(sampleInfo(tSet_old), sampleInfo(tSet_new)))
#    message("The cell slots are not equivalent!")
#
#  if(!assertthat::are_equal(treatmentInfo(tSet_old), treatmentInfo(tSet_new)))
#    message("The drug slots are not equivalent!")
#
#  if(!assertthat::are_equal(treatmentResponse(tSet_old), treatmentResponse(tSet_new)))
#    message("The sensitivity slots are not equivalent!")
#
#  if(!assertthat::are_equal(datasetType(tSet_old), datasetType(tSet_new)))
#    message("The datasetType slots are not equivalent!")
#
#  if(!assertthat::are_equal(tSet_old@perturbation, tSet_new@perturbation))
#    message("The perturbation slots are not equivalent!")
#
# if(!assertthat::are_equal(curation(tSet_old), curation(tSet_new)))
#   message("The curation slots are not equivalent!")
#}

# Utility function to resave all datasets after modifying converttSetMolecularProfiles
#
# Converts all example dastasets specificed as an argument from
#   molecularProfiles as ExpressionSet to molecularProfiles as
#   SummarizedExperiment and saves them in the data folder
#
# @param datasets \code{character} A list of the example datasets to update
#
# @return \code{none} Works by side effects alone to resave all example
#   datasets in a package to have SummarizedExperiments for molecularProfiles
#.resaveAllExampleDatasets <- function(datasets) {
#  for (dataset in datasets) {
#    dataDir <- paste0(grep('data', list.dirs(), value=TRUE))
#    load(paste0(dataDir, '/', dataset, '_old.rda'))
#    assign(dataset, .convertTsetMolecularProfilesToSE(get(dataset)))
#    save(list=dataset, file=paste0(dataDir, '/', dataset, '.rda'), compress='xz')
#  }
#}
#

.eSetToSE <- function(eSet) {
    # Build summarized experiment from eSet
    SE <- SummarizedExperiment::SummarizedExperiment(
        ## TODO:: Do we want to pass an environment for better memory efficiency?
        assays=SimpleList(as.list(Biobase::assayData(eSet))
        ),
        # Switch rearrange columns so that IDs are first, probes second
        rowData=S4Vectors::DataFrame(Biobase::fData(eSet),
                                     rownames=rownames(Biobase::fData(eSet))
        ),
        colData=S4Vectors::DataFrame(Biobase::pData(eSet),
                                     rownames=rownames(Biobase::pData(eSet))
        ),
        metadata=list("experimentData" = eSet@experimentData,
                      "annotation" = Biobase::annotation(eSet),
                      "protocolData" = Biobase::protocolData(eSet)
        )
    )
    ## TODO:: Determine if this can be done in the SE constructor?
    # Extract names from expression set
    SummarizedExperiment::assayNames(SE) <- Biobase::assayDataElementNames(eSet)
    return(SE)
}

#.validateESetToSEConversions <- function(eSet, SE) {
#        for (j in seq_along(assays(SE))) {
#            if(!all(as.list(assayData(eSet))[[j]] ==
#                    as.list(assays(SE))[[j]],
#                    na.rm = TRUE)
#            ) message("The assay data is not equivalent")
#        }
#    ## TODO:: Rewrite this as an apply statement
#        # Checking phenoData
#        if(
#            if (nrow(pData(eSet)) > 0) {
#                !all(
#                    as(eSet@phenoData, value="data.frame") ==
#                    as.data.frame(SE@colData[
#                        seq_len(length(SE@colData) - 1)]),
#                    na.rm = TRUE)
#            } else { FALSE }
#        ) message("The phenoData is not equivalent")
#        # Checking featureData
#        if(
#            if (nrow(fData(eSet)) > 0) {
#                !all(
#                    as(eSet@featureData, value="data.frame") ==
#                    as.data.frame(SE@elementMetadata[
#                        seq_len(length(SE@elementMetadata) - 1)]),
#                    na.rm=TRUE)
#            } else { FALSE }
#        ) message("The featureData is not equivalent")
#        # Checking protocolData
#        if(
#            !all(
#                as(eSet@protocolData, value="data.frame") ==
#                as(SE@metadata$protocolData, value="data.frame"),
#                na.rm = TRUE)
#        ) message("The protocolData is not equivalent")
#
#    if(!assertthat::are_equal(
#        annotation(eSet),
#        metadata(SE)$annotation
#    )) message("The annotation is not equivalent")
#
#    if(!assertthat::are_equal(
#        experimentData(eSet),
#        metadata(SE)$experimentData
#    )) message("The experimentData is not equivalent")
#}