@@ -10,7 +10,7 @@ Maintainer: Dario Strbenac <dario.strbenac@sydney.edu.au>

 VignetteBuilder: knitr

 Encoding: UTF-8

 biocViews: Classification, Survival

-Depends: R (>= 4.1.0), methods, S4Vectors (>= 0.18.0), MultiAssayExperiment (>= 1.6.0), BiocParallel, survival

+Depends: R (>= 4.1.0), generics, methods, S4Vectors (>= 0.18.0), MultiAssayExperiment (>= 1.6.0), BiocParallel, survival

 Imports: grid, genefilter, utils, dplyr, tidyr, rlang, randomForest

 Suggests: limma, edgeR, car, Rmixmod, ggplot2 (>= 3.0.0), gridExtra (>= 2.0.0), cowplot,

         BiocStyle, pamr, PoiClaClu, parathyroidSE, knitr, htmltools, gtable,

@@ -27,7 +27,7 @@ Description: The software formalises a framework for classification in R.

 License: GPL-3

 Packaged: 2014-10-18 11:16:55 UTC; dario

 RoxygenNote: 7.2.1

-SystemRequirements: C++14

+NeedsCompilation: yes

 Collate:

     'ROCplot.R'

     'available.R'

@@ -1,6 +1,11 @@

 # Generated by roxygen2: do not edit by hand

 S3method(predict,trainedByClassifyR)

+S3method(train,DataFrame)

+S3method(train,MultiAssayExperiment)

+S3method(train,data.frame)

+S3method(train,list)

+S3method(train,matrix)

 export(ClassifyResult)

 export(CrossValParams)

 export(FeatureSetCollection)

@@ -36,11 +41,6 @@ export(sampleNames)

 export(samplesMetricMap)

 export(selectionPlot)

 export(totalPredictions)

-export(train.DataFrame)

-export(train.MultiAssayExperiment)

-export(train.data.frame)

-export(train.list)

-export(train.matrix)

 export(tunedParameters)

 exportClasses(ClassifyResult)

 exportClasses(CrossValParams)

@@ -94,6 +94,7 @@ importFrom(S4Vectors,do.call)

 importFrom(S4Vectors,mcols)

 importFrom(dplyr,mutate)

 importFrom(dplyr,n)

+importFrom(generics,train)

 importFrom(rlang,sym)

 importFrom(survival,Surv)

 importFrom(survival,concordance)

@@ -854,14 +854,17 @@ simplifyResults <- function(results, values = c("assay", "classifier", "selectio

 }

 #' @rdname crossValidate

+#' @importFrom generics train

+#' @method train matrix

 #' @export

-train.matrix <-function(x, outcomeTrain, ...)

+train.matrix <- function(x, outcomeTrain, ...)

                {

                  x <- DataFrame(x, check.names = FALSE)

                  train(x, outcomeTrain, ...)

                }

 #' @rdname crossValidate

+#' @method train data.frame

 #' @export

 train.data.frame <- function(x, outcomeTrain, ...)

                     {

@@ -872,20 +875,26 @@ train.data.frame <- function(x, outcomeTrain, ...)

 #' @rdname crossValidate

 #' @param assayIDs A character vector for assays to train with. Special value \code{"all"}

 #' uses all assays in the input object.

+#' @method train DataFrame

 #' @export

 train.DataFrame <- function(x, outcomeTrain, classifier = "randomForest", multiViewMethod = "none", assayIDs = "all", ...) # ... for prepareData.

                    {

               prepArgs <- list(x, outcomeTrain)

               extraInputs <- list(...)

+              prepExtras <- numeric()

               if(length(extraInputs) > 0)

                 prepExtras <- which(names(extrasInputs) %in% .ClassifyRenvir[["prepareDataFormals"]])

               if(length(prepExtras) > 0)

                 prepArgs <- append(prepArgs, extraInputs[prepExtras])

               measurementsAndOutcome <- do.call(prepareData, prepArgs)

+              measurements <- measurementsAndOutcome[["measurements"]]

+              outcomeTrain <- measurementsAndOutcome[["outcome"]]

               classifier <- cleanClassifier(classifier = classifier, measurements = measurements)

-              if(assayIDs == "all") assayIDs <- unique(mcols(x)[, "assay"])

+              if(assayIDs == "all") assayIDs <- unique(mcols(measurements)[, "assay"])

               if(is.null(assayIDs)) assayIDs <- 1

+              names(assayIDs) <- assayIDs

+              names(classifier) <- classifier

               if(multiViewMethod == "none"){

                   resClassifier <-

@@ -897,10 +906,10 @@ train.DataFrame <- function(x, outcomeTrain, classifier = "randomForest", multiV

                                   if(assayIndex != 1) measurementsUse <- measurements[, mcols(measurements)[, "assay"] == assayIndex, drop = FALSE]

                                   classifierParams <- .classifierKeywordToParams(classifierForAssay)

-                                  modellingParams <- ModellingParams(balancing = "none", selectParams = "none",

+                                  modellingParams <- ModellingParams(balancing = "none", selectParams = NULL,

                                                                trainParams = classifierParams$trainParams, predictParams = classifierParams$predictParams)

-                                  .doTrain(measurementsUse, outcomeTrain, NULL, NULL, modellingParams, verbose = 0)[["model"]]

+                                  .doTrain(measurementsUse, outcomeTrain, NULL, NULL, CrossValParams(), modellingParams, verbose = 0)[["model"]]

                                   ## train model

                           },

                           simplify = FALSE)

@@ -908,8 +917,13 @@ train.DataFrame <- function(x, outcomeTrain, classifier = "randomForest", multiV

                       simplify = FALSE)

                   models <- unlist(resClassifier, recursive = FALSE)

-                  names(models) <- assayIDs

-                  class(models) <- c(class(models), "listOfModels")

+                  if(length(models) == 1) {

+                      model <- models[[1]]

+                      class(model) <- c(class(model), "trainedByClassifyR")

+                      models <- NULL

+                  } else {

+                      class(models) <- c(class(models), "listOfModels", "trainedByClassifyR")

+                  }

               }

               ################################

@@ -919,7 +933,8 @@ train.DataFrame <- function(x, outcomeTrain, classifier = "randomForest", multiV

               ### Merging or binding to combine data

               if(multiViewMethod == "merge"){

                   measurementsUse <- measurements[, mcols(measurements)[["assay"]] %in% assayIDs]

-                  .doTrain(measurementsUse, outcomeTrain, NULL, NULL, modellingParams, verbose = 0)[["model"]]

+                  model <- .doTrain(measurementsUse, outcomeTrain, NULL, NULL, crossValParams, modellingParams, verbose = 0)[["model"]]

+                  class(model) <- c("trainedByClassifyR", class(model))

               }

@@ -941,7 +956,8 @@ train.DataFrame <- function(x, outcomeTrain, classifier = "randomForest", multiV

                                     selectParams = NULL,

                                     trainParams = TrainParams(prevalTrainInterface, params = paramsAssays, characteristics = paramsAssays$clinical@trainParams@characteristics),

                                     predictParams = PredictParams(prevalPredictInterface, characteristics = paramsAssays$clinical@predictParams@characteristics))

-                 .doTrain(measurementsUse, outcomeTrain, NULL, NULL, modellingParams, verbose = 0)[["model"]]

+                 model <- .doTrain(measurementsUse, outcomeTrain, NULL, NULL, crossValParams, modellingParams, verbose = 0)[["model"]]

+                 class(model) <- c("trainedByClassifyR", class(model))

               }

               ### Principal Components Analysis to combine data

@@ -956,11 +972,14 @@ train.DataFrame <- function(x, outcomeTrain, classifier = "randomForest", multiV

                 modellingParams <- ModellingParams(balancing = "none", selectParams = NULL,

                                    trainParams = TrainParams(pcaTrainInterface, params = paramsClinical, nFeatures = nFeatures, characteristics = paramsClinical$clinical@trainParams@characteristics),

                                    predictParams = PredictParams(pcaPredictInterface, characteristics = paramsClinical$clinical@predictParams@characteristics))

-                .doTrain(measurementsUse, outcomeTrain, NULL, NULL, modellingParams, verbose = 0)[["model"]]

+                model <- .doTrain(measurementsUse, outcomeTrain, NULL, NULL, crossValParams, modellingParams, verbose = 0)[["model"]]

+                class(model) <- c("trainedByClassifyR", class(model))

               }

+              if(missing(models) || is.null(models)) return(model) else return(models)

           }

 #' @rdname crossValidate

+#' @method train list

 #' @export

 # Each of the first four variables are named lists with names of assays.

 train.list <- function(x, outcomeTrain, ...)

@@ -997,11 +1016,13 @@ train.list <- function(x, outcomeTrain, ...)

 }

 #' @rdname crossValidate

+#' @method train MultiAssayExperiment

 #' @export

 train.MultiAssayExperiment <- function(x, outcomeColumns, ...)

           {

               prepArgs <- list(x, outcomeColumns)

               extraInputs <- list(...)

+              prepExtras <- trainExtras <- numeric()

               if(length(extraInputs) > 0)

                 prepExtras <- which(names(extrasInputs) %in% .ClassifyRenvir[["prepareDataFormals"]])

               if(length(prepExtras) > 0)

@@ -1021,7 +1042,7 @@ train.MultiAssayExperiment <- function(x, outcomeColumns, ...)

 #' \code{DataFrame}, \code{list} (of matrices or data frames) or \code{MultiAssayExperiment} containing

 #' the data to make predictions with with either a fitted model created by \code{train} or the final model

 #' stored in a \code{\link{ClassifyResult}} object.

-

+#' @method predict trainedByClassifyR

 #' @export

 predict.trainedByClassifyR <- function(object, newData, ...)

 {

@@ -191,7 +191,7 @@ input data. Autmomatically reducing to smaller number.")

   }

   # Some classifiers have one function for training and testing, so that's why test data is also passed in.

-  trained <- tryCatch(.doTrain(measurementsTrain, outcomeTrain, measurementsTest, outcomeTest, modellingParams, verbose),

+  trained <- tryCatch(.doTrain(measurementsTrain, outcomeTrain, measurementsTest, outcomeTest, crossValParams, modellingParams, verbose),

                       error = function(error) error[["message"]])

   if(is.character(trained)) return(trained) # An error occurred.

@@ -238,7 +238,7 @@ input data. Autmomatically reducing to smaller number.")

     {

       measurementsTrainLess1 <- measurementsTrain[, -selectedIndex, drop = FALSE]

       measurementsTestLess1 <- measurementsTest[, -selectedIndex, drop = FALSE]

-      modelWithoutOne <- tryCatch(.doTrain(measurementsTrainLess1, outcomeTrain, measurementsTestLess1, outcomeTest, modellingParams, verbose),

+      modelWithoutOne <- tryCatch(.doTrain(measurementsTrainLess1, outcomeTrain, measurementsTestLess1, outcomeTest, crossValParams, modellingParams, verbose),

                                   error = function(error) error[["message"]])

       if(!is.null(modellingParams@predictParams))

       predictedOutcomeWithoutOne <- tryCatch(.doTest(modelWithoutOne[["model"]], measurementsTestLess1, modellingParams@predictParams, verbose),

@@ -1,6 +1,6 @@

 # Random Forest

 RFparams <- function() {

-    trainParams <- TrainParams(randomForestTrainInterface, tuneParams = list(mTryProportion = c(0.25, 0.33, 0.50, 0.66, 0.75, 1.00), ntree = seq(100, 500, 100)),

+    trainParams <- TrainParams(randomForestTrainInterface, tuneParams = list(mTryProportion = c(0.25, 0.33, 0.50, 0.66, 0.75, 1.00), ntree = seq(100, 500, 100), performanceType = "Balanced Error"),

                                getFeatures = forestFeatures)

     predictParams <- PredictParams(randomForestPredictInterface)

@@ -9,7 +9,7 @@ RFparams <- function() {

 # Random Survival Forest

 RSFparams <- function() {

-    trainParams <- TrainParams(rfsrcTrainInterface, tuneParams = list(mTryProportion = c(0.25, 0.33, 0.50, 0.66, 0.75, 1.00), ntree = seq(100, 500, 100)))

+    trainParams <- TrainParams(rfsrcTrainInterface, tuneParams = list(mTryProportion = c(0.25, 0.33, 0.50, 0.66, 0.75, 1.00), ntree = seq(100, 500, 100), performanceType = "Balanced Error"))

     predictParams <- PredictParams(rfsrcPredictInterface)

     return(list(trainParams = trainParams, predictParams = predictParams))

@@ -17,7 +17,7 @@ RSFparams <- function() {

 # k Nearest Neighbours

 kNNparams <- function() {

-    trainParams <- TrainParams(kNNinterface, tuneParams = list(k = 1:5))

+    trainParams <- TrainParams(kNNinterface, tuneParams = list(k = 1:5, performanceType = "Balanced Error"))

     predictParams <- NULL

     return(list(trainParams = trainParams, predictParams = predictParams))

 }

@@ -40,7 +40,7 @@ elasticNetGLMparams <- function() {

 # Support Vector Machine

 SVMparams = function() {

-    trainParams <- TrainParams(SVMtrainInterface, tuneParams = list(kernel = c("linear", "polynomial", "radial", "sigmoid"), cost = 10^(-3:3)))

+    trainParams <- TrainParams(SVMtrainInterface, tuneParams = list(kernel = c("linear", "polynomial", "radial", "sigmoid"), cost = 10^(-3:3), performanceType = "Balanced Error"))

     predictParams <- PredictParams(SVMpredictInterface)

     return(list(trainParams = trainParams, predictParams = predictParams))

@@ -64,7 +64,7 @@ DLDAparams = function() {

 # naive Bayes Kernel

 naiveBayesParams <- function() {

-    trainParams <- TrainParams(naiveBayesKernel, tuneParams = list(difference = c("unweighted", "weighted")))

+    trainParams <- TrainParams(naiveBayesKernel, tuneParams = list(difference = c("unweighted", "weighted"), performanceType = "Balanced Error"))

     predictParams <- NULL

     return(list(trainParams = trainParams, predictParams = predictParams))

 }

@@ -106,7 +106,6 @@

   tuneParams <- modellingParams@selectParams@tuneParams

   performanceType <- tuneParams[["performanceType"]]

   topNfeatures <- tuneParams[["nFeatures"]]

-  tuneMode <- ifelse("tuneMode" %in% names(tuneParams), tuneParams[["tuneMode"]], crossValParams@tuneMode)

   tuneParams <- tuneParams[-match(c("performanceType", "nFeatures"), names(tuneParams))] # Only used as evaluation metric.

   # Make selectParams NULL, since we are currently doing selection and it shouldn't call

@@ -136,7 +135,7 @@

     if(attr(featureRanking, "name") == "previousSelection") # Actually selection not ranking.

       return(list(NULL, rankings[[1]], NULL))

-    if(tuneMode == "none") # No parameters to choose between.

+    if(crossValParams@tuneMode == "none") # No parameters to choose between.

         return(list(NULL, rankings[[1]], NULL))

     tuneParamsTrain <- list(topN = topNfeatures)

@@ -257,7 +256,7 @@

 # Code to create a function call to a training function. Might also do training and testing

 # within the same function, so test samples are also passed in case they are needed.

-.doTrain <- function(measurementsTrain, outcomeTrain, measurementsTest, outcomeTest, modellingParams, verbose)

+.doTrain <- function(measurementsTrain, outcomeTrain, measurementsTest, outcomeTest, crossValParams, modellingParams, verbose)

 {

   tuneDetails <- NULL

   if(!is.null(modellingParams@trainParams@tuneParams) && is.null(modellingParams@selectParams))

@@ -278,10 +277,10 @@

         predictions <- result[["predictions"]]

         if(class(predictions) == "data.frame")

-          predictedOutcome <- predictions[, "outcome"]

+          predictedOutcome <- predictions[, colnames(predictions) %in% c("class", "risk")]

         else

           predictedOutcome <- predictions

-        calcExternalPerformance(outcomeTest, predictedOutcome, performanceType)

+        calcExternalPerformance(outcomeTrain, predictedOutcome, performanceType)

       } else {

         result <- runTests(measurementsTrain, outcomeTrain,

                            crossValParams, modellingParams,

@@ -98,11 +98,11 @@ crossValidate(measurements, outcome, ...)

   ...

 )

-train.matrix(x, outcomeTrain, ...)

+\method{train}{matrix}(x, outcomeTrain, ...)

-train.data.frame(x, outcomeTrain, ...)

+\method{train}{data.frame}(x, outcomeTrain, ...)

-train.DataFrame(

+\method{train}{DataFrame}(

   x,

   outcomeTrain,

   classifier = "randomForest",

@@ -111,9 +111,9 @@ train.DataFrame(

   ...

 )

-train.list(x, outcomeTrain, ...)

+\method{train}{list}(x, outcomeTrain, ...)

-train.MultiAssayExperiment(x, outcomeColumns, ...)

+\method{train}{MultiAssayExperiment}(x, outcomeColumns, ...)

 \method{predict}{trainedByClassifyR}(object, newData, ...)

 }

deleted file mode 100644

Binary files a/src/coxformatrices.o and /dev/null differ