make caret and ranger suggested packages

NAMESPACE

@@ -307,8 +307,6 @@ importFrom(ape,as.phylo)
 importFrom(ape,drop.tip)
 importFrom(ape,nodelabels)
 importFrom(ape,plot.phylo)
-importFrom(caret,train)
-importFrom(caret,trainControl)
 importFrom(cluster,clara)
 importFrom(cowplot,get_legend)
 importFrom(cowplot,plot_grid)
@@ -375,7 +373,6 @@ importFrom(pbapply,pbapply)
 importFrom(pbapply,pblapply)
 importFrom(pbapply,pbsapply)
 importFrom(png,readPNG)
-importFrom(ranger,ranger)
 importFrom(reshape2,melt)
 importFrom(reticulate,dict)
 importFrom(reticulate,import)

R/cluster_determination.R

@@ -334,7 +334,6 @@ NumberClusters <- function(object) {
 #'
 #' @import Matrix
 #' @importFrom stats predict
-#' @importFrom ranger ranger
 #'
 #' @export
 #'
@@ -357,6 +356,7 @@ ClassifyCells <- function(
   new.data = NULL,
   ...
 ) {
+  PackageCheck('ranger')
   # build the classifier
   if (missing(classifier)){
     classifier <- BuildRFClassifier(
@@ -398,7 +398,6 @@ ClassifyCells <- function(
 #' @return Returns the random forest classifier
 #'
 #' @import Matrix
-#' @importFrom ranger ranger
 #'
 #' @export
 #'
@@ -431,7 +430,7 @@ BuildRFClassifier <- function(
   if (verbose) {
     message("Training Classifier ...")
   }
-  classifier <- ranger(
+  classifier <- ranger::ranger(
     data = training.data,
     dependent.variable.name = "class",
     classification = TRUE,

R/cluster_validation.R

@@ -13,7 +13,6 @@ NULL
 #' clusters
 #' @param acc.cutoff Accuracy cutoff for classifier
 #' @param verbose Controls whether to display progress and merging results
-#' @importFrom caret trainControl train
 #' @return Returns a Seurat object, object@@ident has been updated with new
 #' cluster info
 #' @export
@@ -35,6 +34,7 @@ ValidateClusters <- function(
   acc.cutoff = 0.9,
   verbose = TRUE
 ) {
+  PackageCheck('caret')
   # probably should refactor to make cleaner
   if (length(x = object@snn) > 1) {
     SNN.use <- object@snn
@@ -133,7 +133,6 @@ ValidateClusters <- function(
 #' @param pc.use Which PCs to use for model construction
 #' @param top.genes Use the top X genes for model construction
 #' @param acc.cutoff Accuracy cutoff for classifier
-#' @importFrom caret trainControl train
 #' @return Returns a Seurat object, object@@ident has been updated with
 #' new cluster info
 #' @export
@@ -155,6 +154,7 @@ ValidateSpecificClusters <- function(
   top.genes = 30,
   acc.cutoff = 0.9
 ) {
+  PackageCheck('caret')
   acc <- RunClassifier(
     object = object,
     group1 = cluster1,
@@ -203,9 +203,9 @@ RunClassifier <- function(object, group1, group2, pcs, num.genes) {
   xv <- apply(X = x, MARGIN = 2, FUN = var)
   x  <- x[, names(x = which(xv > 0))]
   # run k-fold cross validation
-  ctrl <- trainControl(method = "repeatedcv", repeats = 5)
+  ctrl <- caret::trainControl(method = "repeatedcv", repeats = 5)
   set.seed(seed = 1500)
-  model <- train(
+  model <- caret::train(
     x = x,
     y = as.factor(x = y),
     formula = as.factor(x = y) ~ .,