@@ -30,6 +30,8 @@ Imports:

     e1071, 

     ape, 

     kernlab,

+    tools,

+    BiocFileCache,

     utils

 Suggests: 

     knitr,

@@ -39,7 +41,6 @@ Suggests:

 VignetteBuilder: knitr

 Depends: R (>= 4.1), Seurat, SingleCellExperiment, SummarizedExperiment

 LazyData: true

-LazyDataCompression: xz

 RoxygenNote: 7.1.1

 URL: https://github.com/grisslab/scAnnotatR

 BugReports: https://github.com/grisslab/scAnnotatR/issues/new

@@ -6,6 +6,7 @@ export(caret_model)

 export(cell_type)

 export(classify_cells)

 export(delete_model)

+export(load_models)

 export(marker_genes)

 export(p_thres)

 export(parent)

@@ -16,6 +17,7 @@ export(scAnnotatR)

 export(test_classifier)

 export(train_classifier)

 exportMethods(show)

+import(BiocFileCache)

 import(ROCR)

 import(SingleCellExperiment)

 import(ape)

@@ -27,6 +29,7 @@ import(ggplot2)

 import(kernlab, except = c(alpha, predict))

 import(methods)

 import(pROC)

+import(tools)

 importFrom(Seurat,GetAssayData)

 importFrom(SummarizedExperiment,assay)

 importFrom(SummarizedExperiment,colData)

@@ -132,6 +132,79 @@ setMethod("train_classifier", c("train_obj" = "Seurat"),

   return(object)

 })

+#' @inherit train_classifier

+#' 

+#' @param sce_tag_slot string, name of annotation slot indicating 

+#' cell tag/label in the training object.

+#' For \code{\link{SingleCellExperiment}} object, default value is "ident".  

+#' Expected values are string (A-Z, a-z, 0-9, no special character accepted) 

+#' or binary/logical, 0/"no"/F/FALSE: not being new cell type, 

+#' 1/"yes"/T/TRUE: being new cell type.

+#' @param sce_parent_tag_slot string, name of a slot in cell meta data 

+#' indicating pre-assigned/predicted cell type. 

+#' Default field is "predicted_cell_type".

+#' This field would have been filled automatically 

+#' when user called classify_cells function. 

+#' The slot must contain only string values. 

+#' @param sce_assay name of assay to use in training object. 

+#' Default to 'logcounts' assay.

+#' 

+#' @import SingleCellExperiment

+#' @importFrom SummarizedExperiment assay

+#' 

+#' @rdname train_classifier

+setMethod("train_classifier", c("train_obj" = "SingleCellExperiment"), 

+          function(train_obj, cell_type, marker_genes, parent_cell = NA_character_,

+                   parent_classifier = NULL, path_to_models = "default", 

+                   zscore = TRUE, sce_tag_slot = "ident", 

+                   sce_parent_tag_slot = "predicted_cell_type", 

+                   sce_assay = 'logcounts', ...) {

+  # solve duplication of cell names

+  colnames(train_obj) <- make.unique(colnames(train_obj), sep = '_')

+  

+  # convert Seurat object to matrix

+  mat = SummarizedExperiment::assay(train_obj, sce_assay)

+  

+  tag = SummarizedExperiment::colData(train_obj)[, sce_tag_slot]

+  names(tag) <- colnames(train_obj)

+  

+  if (sce_parent_tag_slot %in% colnames(SummarizedExperiment::colData(train_obj))) {

+    parent_tag <- SummarizedExperiment::colData(train_obj)[, sce_parent_tag_slot]

+    names(parent_tag) <- colnames(train_obj)

+  } else parent_tag <- NULL

+  

+  object <- train_classifier_func(mat, tag, cell_type, marker_genes, 

+                                  parent_tag, parent_cell, parent_classifier,

+                                  path_to_models, zscore)

+  

+  return(object)

+})

+

+#' Train cell type from matrix

+#' 

+#' @description Train a classifier for a new cell type from expression matrix

+#' and tag 

+#' If cell type has a parent, only available for \code{\link{scAnnotatR}}

+#' object as parent cell classifying model.

+#' @param mat expression matrix of size n x m, n: genes, m: cells

+#' @param tag named list indicating cell label

+#' @param cell_type string indicating the name of the subtype

+#' This must exactly match cell tag/label if cell tag/label is a string.

+#' @param parent_tag named list indicating parent cell type

+#' @param marker_genes list of marker genes used for the new training model

+#' @param parent_cell string indicated the name of the parent cell type, 

+#' if parent cell type classifier has already been saved in model database.

+#' Adjust path_to_models for exact database.  

+#' @param parent_classifier classification model for the parent cell type

+#' @param path_to_models path to the folder containing the model database. 

+#' As default, the pretrained models in the package will be used. 

+#' If user has trained new models, indicate the folder containing the 

+#' new_models.rda file.

+#' @param zscore whether gene expression in train_obj is transformed to zscore

+#' 

+#' @return caret trained model

+#' 

+#' @rdname internal

 train_classifier_func <- function(mat, tag, cell_type, marker_genes, 

                                   parent_tag, parent_cell, parent_classifier, 

                                   path_to_models, zscore) {

@@ -139,7 +212,7 @@ train_classifier_func <- function(mat, tag, cell_type, marker_genes,

   processed_parent <- process_parent_classifier(

     mat, parent_tag, parent_cell, parent_classifier, path_to_models, zscore

   )

-

+  

   # check parent-child coherence

   if (!is.null(processed_parent$pos_parent)) {

     tag <- check_parent_child_coherence(

@@ -206,54 +279,6 @@ train_classifier_func <- function(mat, tag, cell_type, marker_genes,

   return(object)

 }

-#' @inherit train_classifier

-#' 

-#' @param sce_tag_slot string, name of annotation slot indicating 

-#' cell tag/label in the training object.

-#' For \code{\link{SingleCellExperiment}} object, default value is "ident".  

-#' Expected values are string (A-Z, a-z, 0-9, no special character accepted) 

-#' or binary/logical, 0/"no"/F/FALSE: not being new cell type, 

-#' 1/"yes"/T/TRUE: being new cell type.

-#' @param sce_parent_tag_slot string, name of a slot in cell meta data 

-#' indicating pre-assigned/predicted cell type. 

-#' Default field is "predicted_cell_type".

-#' This field would have been filled automatically 

-#' when user called classify_cells function. 

-#' The slot must contain only string values. 

-#' @param sce_assay name of assay to use in training object. 

-#' Default to 'logcounts' assay.

-#' 

-#' @import SingleCellExperiment

-#' @importFrom SummarizedExperiment assay

-#' 

-#' @rdname train_classifier

-setMethod("train_classifier", c("train_obj" = "SingleCellExperiment"), 

-          function(train_obj, cell_type, marker_genes, parent_cell = NA_character_,

-                   parent_classifier = NULL, path_to_models = "default", 

-                   zscore = TRUE, sce_tag_slot = "ident", 

-                   sce_parent_tag_slot = "predicted_cell_type", 

-                   sce_assay = 'logcounts', ...) {

-  # solve duplication of cell names

-  colnames(train_obj) <- make.unique(colnames(train_obj), sep = '_')

-  

-  # convert Seurat object to matrix

-  mat = SummarizedExperiment::assay(train_obj, sce_assay)

-  

-  tag = SummarizedExperiment::colData(train_obj)[, sce_tag_slot]

-  names(tag) <- colnames(train_obj)

-  

-  if (sce_parent_tag_slot %in% colnames(SummarizedExperiment::colData(train_obj))) {

-    parent_tag <- SummarizedExperiment::colData(train_obj)[, sce_parent_tag_slot]

-    names(parent_tag) <- colnames(train_obj)

-  } else parent_tag <- NULL

-  

-  object <- train_classifier_func(mat, tag, cell_type, marker_genes, 

-                                  parent_tag, parent_cell, parent_classifier,

-                                  path_to_models, zscore)

-  

-  return(object)

-})

-

 #' Testing process.

 #' 

 #' @description Testing process. 

@@ -363,6 +388,76 @@ setMethod("test_classifier", c("test_obj" = "Seurat",

   return(return_val)

 })

+#' @inherit test_classifier

+#' 

+#' @param sce_tag_slot string, name of annotation slot 

+#' indicating cell tag/label in the testing object.

+#' Strings indicating cell types are expected in this slot. 

+#' Default value is "ident".  

+#' Expected values are string (A-Z, a-z, 0-9, no special character accepted) 

+#' or binary/logical, 0/"no"/F/FALSE: not being new cell type, 

+#' 1/"yes"/T/TRUE: being new cell type.

+#' @param sce_parent_tag_slot string, name of tag slot in cell meta data

+#' indicating pre-assigned/predicted parent cell type. 

+#' Default is "predicted_cell_type".

+#' The slot must contain only string values. 

+#' @param sce_assay name of assay to use in \code{\link{SingleCellExperiment}}

+#' object, defaults to 'logcounts' assay.

+#'  

+#' @import SingleCellExperiment

+#' @importFrom SummarizedExperiment assay

+#' 

+#' @rdname test_classifier

+setMethod("test_classifier", c("test_obj" = "SingleCellExperiment", 

+                               "classifier" = "scAnnotatR"), 

+          function(test_obj, classifier, target_cell_type = NULL, 

+                   parent_classifier = NULL, path_to_models = "default", 

+                   zscore = TRUE, sce_tag_slot = "ident", 

+                   sce_parent_tag_slot = "predicted_cell_type", 

+                   sce_assay = 'logcounts', ...) {

+  # solve duplication of cell names

+  colnames(test_obj) <- make.unique(colnames(test_obj), sep = '_')

+  . <- fpr <- tpr <- NULL

+  

+  # convert SCE object to matrix

+  mat = SummarizedExperiment::assay(test_obj, sce_assay)

+  

+  tag = SummarizedExperiment::colData(test_obj)[, sce_tag_slot]

+  names(tag) <- colnames(test_obj)

+  

+  if (sce_parent_tag_slot %in% colnames(SummarizedExperiment::colData(test_obj))) {

+    parent_tag <- SummarizedExperiment::colData(test_obj)[, sce_parent_tag_slot]

+    names(parent_tag) <- colnames(test_obj)

+  } else parent_tag <- NULL

+  

+  return_val <- test_classifier_func(mat, tag, classifier, parent_tag,

+                                     target_cell_type, parent_classifier,

+                                     path_to_models, zscore)

+  

+  return(return_val)

+})

+

+#' Run testing process from matrix and tag

+#' 

+#' @description Testing process from matrix and tag 

+#' @param mat expression matrix of size n x m, n: genes, m: cells

+#' @param tag named list indicating cell label

+#' @param classifier classification model

+#' @param parent_tag named list indicating parent cell type

+#' @param target_cell_type vector indicating other cell types than cell labels 

+#' that can be considered as the main cell type in classifier, 

+#' for example, c("plasma cell", "b cell", "b cells", "activating b cell"). 

+#' Default as NULL.

+#' @param parent_classifier classification model for the parent cell type

+#' @param path_to_models path to the folder containing the model database. 

+#' As default, the pretrained models in the package will be used. 

+#' If user has trained new models, indicate the folder containing the 

+#' new_models.rda file.

+#' @param zscore whether gene expression in train_obj is transformed to zscore

+#' 

+#' @return model performance statistics

+#' 

+#' @rdname internal

 test_classifier_func <- function(mat, tag, classifier, parent_tag, 

                                  target_cell_type, parent_classifier,

                                  path_to_models, zscore) {

@@ -415,55 +510,6 @@ test_classifier_func <- function(mat, tag, classifier, parent_tag,

   return(return_val)

 }

-#' @inherit test_classifier

-#' 

-#' @param sce_tag_slot string, name of annotation slot 

-#' indicating cell tag/label in the testing object.

-#' Strings indicating cell types are expected in this slot. 

-#' Default value is "ident".  

-#' Expected values are string (A-Z, a-z, 0-9, no special character accepted) 

-#' or binary/logical, 0/"no"/F/FALSE: not being new cell type, 

-#' 1/"yes"/T/TRUE: being new cell type.

-#' @param sce_parent_tag_slot string, name of tag slot in cell meta data

-#' indicating pre-assigned/predicted parent cell type. 

-#' Default is "predicted_cell_type".

-#' The slot must contain only string values. 

-#' @param sce_assay name of assay to use in \code{\link{SingleCellExperiment}}

-#' object, defaults to 'logcounts' assay.

-#'  

-#' @import SingleCellExperiment

-#' @importFrom SummarizedExperiment assay

-#' 

-#' @rdname test_classifier

-setMethod("test_classifier", c("test_obj" = "SingleCellExperiment", 

-                               "classifier" = "scAnnotatR"), 

-          function(test_obj, classifier, target_cell_type = NULL, 

-                   parent_classifier = NULL, path_to_models = "default", 

-                   zscore = TRUE, sce_tag_slot = "ident", 

-                   sce_parent_tag_slot = "predicted_cell_type", 

-                   sce_assay = 'logcounts', ...) {

-  # solve duplication of cell names

-  colnames(test_obj) <- make.unique(colnames(test_obj), sep = '_')

-  . <- fpr <- tpr <- NULL

-  

-  # convert SCE object to matrix

-  mat = SummarizedExperiment::assay(test_obj, sce_assay)

-  

-  tag = SummarizedExperiment::colData(test_obj)[, sce_tag_slot]

-  names(tag) <- colnames(test_obj)

-  

-  if (sce_parent_tag_slot %in% colnames(SummarizedExperiment::colData(test_obj))) {

-    parent_tag <- SummarizedExperiment::colData(test_obj)[, sce_parent_tag_slot]

-    names(parent_tag) <- colnames(test_obj)

-  } else parent_tag <- NULL

-  

-  return_val <- test_classifier_func(mat, tag, classifier, parent_tag,

-                                     target_cell_type, parent_classifier,

-                                     path_to_models, zscore)

-  

-  return(return_val)

-})

-

 #' Plot roc curve

 #' 

 #' @param test_result result of test_classifier function

@@ -8,16 +8,4 @@

 #' @source WEIZMANN INSTITUTE OF SCIENCE

 #' @author Itay Tirosh, 2016-04-05

 #' @keywords datasets

-"tirosh_mel80_example"

-

-#' @name default_models

-#' @title Pretrained classifiers for human cells

-#' @description Pretrained classifier obtained by training and testing on the 

-#' Sade-Feldman melanoma dataset, the Jerby-Arnon melanoma dataset, the Haniffa

-#' Skin Cell Atlas and the Haniffa Covid-19 Cell Atlas.

-#' @docType data

-#' @usage default_models

-#' @format a list of \code{\link{scAnnotatR}} objects

-#' @author Vy Nguyen, June 2021

-#' @keywords datasets

-"default_models"

\ No newline at end of file

+"tirosh_mel80_example"

\ No newline at end of file

@@ -107,14 +107,16 @@ transform_to_zscore <- function(mat) {

 #' @return list of classifiers

 #' 

 #' @importFrom utils data

-#' @rdname internal

+#' @rdname load_models

+#' @export

 load_models <- function(path_to_models) {

   # prevents R CMD check note

   model_list <- NULL

   data_env <- new.env(parent = emptyenv())

   if (path_to_models == "default") {

-    utils::data("default_models", envir = data_env)

+    models_path <- download_data_file(TRUE) # more function: if user want to save cache

+    load(models_path, envir = data_env)

     model_list <- data_env[["default_models"]]

   } else {

     models_path <- file.path(path_to_models, "new_models.rda")

@@ -608,4 +610,43 @@ classify_clust <- function(clusts, most_probable_cell_type) {

   names(clust.pred) <- levels(clusts)

   converted_pred <- unlist(lapply(clusts, function(x) clust.pred[[x]]))

   return(converted_pred)

-}

\ No newline at end of file

+}

+

+#' Create a BiocFileCache object

+#'

+#' @return BiocFileCache object 

+#' @import tools

+#' @import BiocFileCache

+#' 

+#' @rdname internal

+.get_cache <-

+  function()

+  {

+    cache <- tools::R_user_dir("scAnnotatR", which="cache")

+    BiocFileCache::BiocFileCache(cache)

+  }

+

+#' Download and store default models in cache

+#' @param verbose logical indicating downloading the file or not 

+#'

+#' @return path to the downloaded file in cache 

+#' @import BiocFileCache

+#' 

+#' @rdname internal

+download_data_file <-

+  function(verbose = FALSE)

+  {

+    fileURL <- "https://github.com/grisslab/scAnnotatR-models/blob/main/default_models.rda?raw=true"

+    

+    bfc <- .get_cache()

+    rid <- BiocFileCache::bfcquery(bfc, "default_models", "rname")$rid

+    if (!length(rid)) {

+      if (verbose)

+        message("Downloading default models..." )

+      rid <- names(BiocFileCache::bfcadd(bfc, "default_models", fileURL))

+    }

+    if (isFALSE(BiocFileCache::bfcneedsupdate(bfc, rid)))

+      BiocFileCache::bfcdownload(bfc, rid)

+    

+    BiocFileCache::bfcrpath(bfc, rids = rid)

+  }

\ No newline at end of file

@@ -9,7 +9,7 @@

 #' both of them once. In addition, default pretrained models

 #' of the package cannot be changed or removed. 

 #' This can be done with the new trained model list.

-#' @param path.to.models path to the folder containing the list of new models.

+#' @param path_to_models path to the folder containing the list of new models.

 #' 

 #' @return no return value, but the model is now saved to database

 #' 

@@ -27,7 +27,7 @@

 #' 

 #' # save the trained classifier to system 

 #' # test classifier can be used before this step

-#' save_new_model(new_model = classifier_t, path.to.models = tempdir())

+#' save_new_model(new_model = classifier_t, path_to_models = tempdir())

 #' 

 #' # verify if new model has been saved

 #' print(names(load(file.path(tempdir(), "new_models.rda"))))

@@ -35,20 +35,24 @@

 #' 

 #' @export

 save_new_model <- function(new_model, include.default = TRUE, 

-                    path.to.models = tempdir()) {

+                    path_to_models = tempdir()) {

   default_models <- NULL

+  data_env <- new.env(parent = emptyenv())

-  utils::data("default_models")

-  new_models.file.path = file.path(path.to.models, "new_models.rda")

+  new_models.file.path = file.path(path_to_models, "new_models.rda")

   if (file.exists(new_models.file.path)) {

-    load(new_models.file.path)

+    load(new_models.file.path, data_env)

+    new_models <- data_env[["new_models"]]

   } else {

     new_models = NULL

   }

   if (include.default == TRUE) {

     # default models not in new_models will be added to new_models

+    path_to_default_models <- download_data_file(TRUE)

+    load(path_to_default_models, envir = data_env)

+    default_models <- data_env[["default_models"]]

     to.be.added <- default_models[!names(default_models)%in%names(new_models)]

     new_models <- append(to.be.added, new_models)

   }

@@ -77,7 +81,7 @@ save_new_model <- function(new_model, include.default = TRUE,

 #' Plant tree from list of models

 #' 

-#' @param models.file.path list of models. If not provided, 

+#' @param path_to_models list of models. If not provided, 

 #' list of default pretrained models in the package will be used.

 #'  

 #' @return tree structure and plot of tree 

@@ -92,24 +96,12 @@ save_new_model <- function(new_model, include.default = TRUE,

 #' plant_tree()

 #' 

 #' @export

-plant_tree <- function(models.file.path = "default") { 

+plant_tree <- function(path_to_models = "default") { 

   data_env <- new.env(parent = emptyenv())

   root.name <- "cell types"

-  if (models.file.path == "default") {

-    utils::data("default_models", envir = data_env)

-    model_list <- data_env[['default_models']]

-  } else {

-    models_file_path <- file.path(models.file.path, "new_models.rda")

-    if (!file.exists(models_file_path)) {

-      stop("No file exists in the indicated models file path", 

-           call. = FALSE)

-    } else {

-      load(models_file_path, envir = data_env)

-      model_list <- data_env[['new_models']]

-    }

-  }

+  model_list <- load_models(path_to_models) 

   tree <- NULL

   if (!is.null(model_list)) {

@@ -146,7 +138,7 @@ plant_tree <- function(models.file.path = "default") {

 #' @param cell_type string indicating the cell type of which 

 #' the model will be removed from package

 #' Attention: deletion of a parent model will also delete all of child model.

-#' @param path.to.models path to the folder containing 

+#' @param path_to_models path to the folder containing 

 #' the list of models in which the to-be-deleted model is.

 #' 

 #' @return no return value, but the model is deleted from database

@@ -162,16 +154,19 @@ plant_tree <- function(models.file.path = "default") {

 #' marker_genes = selected_marker_genes_T, cell_type = "t cells")

 #' 

 #' # save a classifier to system

-#' save_new_model(new_model = classifier_t, path.to.models = tempdir())

+#' save_new_model(new_model = classifier_t, path_to_models = tempdir())

 #' 

 #' # delete classifier from system

-#' delete_model("t cells", path.to.models = tempdir())

+#' delete_model("t cells", path_to_models = tempdir())

 #' @export

-delete_model <- function(cell_type, path.to.models = tempdir()) {

+delete_model <- function(cell_type, path_to_models = tempdir()) {

   new_models <- NULL

   data_env <- new.env(parent = emptyenv())

-  new_models.file.path <- file.path(path.to.models, "new_models.rda")

+  if (path_to_models == 'default')

+    stop("Cannot delete default models.", call. = FALSE)

+  

+  new_models.file.path <- file.path(path_to_models, "new_models.rda")

   if (!file.exists(new_models.file.path)) {

     stop("No list of models available", call. = FALSE)

   } else {

deleted file mode 100644

Binary files a/data/default_models.rda and /dev/null differ

deleted file mode 100644

@@ -1,21 +0,0 @@

-% Generated by roxygen2: do not edit by hand

-% Please edit documentation in R/data.R

-\docType{data}

-\name{default_models}

-\alias{default_models}

-\title{Pretrained classifiers for human cells}

-\format{

-a list of \code{\link{scAnnotatR}} objects

-}

-\usage{

-default_models

-}

-\description{

-Pretrained classifier obtained by training and testing on the 

-Sade-Feldman melanoma dataset, the Jerby-Arnon melanoma dataset, the Haniffa

-Skin Cell Atlas and the Haniffa Covid-19 Cell Atlas.

-}

-\author{

-Vy Nguyen, June 2021

-}

-\keyword{datasets}

@@ -4,14 +4,14 @@

 \alias{delete_model}

 \title{Delete model/branch from package}

 \usage{

-delete_model(cell_type, path.to.models = tempdir())

+delete_model(cell_type, path_to_models = tempdir())

 }

 \arguments{

 \item{cell_type}{string indicating the cell type of which 

 the model will be removed from package

 Attention: deletion of a parent model will also delete all of child model.}

-\item{path.to.models}{path to the folder containing 

+\item{path_to_models}{path to the folder containing 

 the list of models in which the to-be-deleted model is.}

 }

 \value{

@@ -31,8 +31,8 @@ classifier_t <- train_classifier(train_obj = tirosh_mel80_example,

 marker_genes = selected_marker_genes_T, cell_type = "t cells")

 # save a classifier to system

-save_new_model(new_model = classifier_t, path.to.models = tempdir())

+save_new_model(new_model = classifier_t, path_to_models = tempdir())

 # delete classifier from system

-delete_model("t cells", path.to.models = tempdir())

+delete_model("t cells", path_to_models = tempdir())

 }

@@ -1,5 +1,5 @@

 % Generated by roxygen2: do not edit by hand

-% Please edit documentation in R/class.R, R/support.R

+% Please edit documentation in R/class.R, R/classifier.R, R/support.R

 \name{checkObjectValidity}

 \alias{checkObjectValidity}

 \alias{checkCellTypeValidity}

@@ -13,10 +13,11 @@

 \alias{caret_model<-,scAnnotatR-method}

 \alias{marker_genes<-}

 \alias{marker_genes<-,scAnnotatR-method}

+\alias{train_classifier_func}

+\alias{test_classifier_func}

 \alias{balance_dataset}

 \alias{train_func}

 \alias{transform_to_zscore}

-\alias{load_models}

 \alias{select_marker_genes}

 \alias{check_parent_child_coherence}

 \alias{check_parent_child_coherence,dgCMatrix,vector-method}

@@ -31,6 +32,8 @@

 \alias{verify_parent}

 \alias{test_performance}

 \alias{classify_clust}

+\alias{.get_cache}

+\alias{download_data_file}

 \title{Internal functions of scAnnotatR package}

 \usage{

 checkObjectValidity(object)

@@ -57,14 +60,35 @@ marker_genes(classifier) <- value

 \S4method{marker_genes}{scAnnotatR}(classifier) <- value

+train_classifier_func(

+  mat,

+  tag,

+  cell_type,

+  marker_genes,

+  parent_tag,

+  parent_cell,

+  parent_classifier,

+  path_to_models,

+  zscore

+)

+

+test_classifier_func(

+  mat,

+  tag,

+  classifier,

+  parent_tag,

+  target_cell_type,

+  parent_classifier,

+  path_to_models,

+  zscore

+)

+

 balance_dataset(mat, tag)

 train_func(mat, tag)

 transform_to_zscore(mat)

-load_models(path_to_models)

-

 select_marker_genes(mat, marker_genes)

 check_parent_child_coherence(

@@ -120,6 +144,10 @@ verify_parent(mat, classifier, meta.data)

 test_performance(mat, classifier, tag)

 classify_clust(clusts, most_probable_cell_type)

+

+.get_cache()

+

+download_data_file(verbose = FALSE)

 }

 \arguments{

 \item{object}{The request classifier to check.}

@@ -142,25 +170,25 @@ classify_clust(clusts, most_probable_cell_type)

 \item{tag}{tag of data}

-\item{path_to_models}{path to databases, or by default}

-

-\item{pos_parent}{a vector indicating parent classifier prediction}

-

-\item{parent_cell}{name of parent cell type}

-

-\item{target_cell_type}{alternative cell types (in case of testing classifier)}

-

 \item{parent_tag}{vector, named list indicating pre-assigned/predicted 

 parent cell type}

-\item{parent_cell_type}{name of parent cell type}

+\item{parent_cell}{name of parent cell type}

 \item{parent_classifier}{\code{\link{scAnnotatR}} object corresponding 

 to classification model for the parent cell type}

+\item{path_to_models}{path to databases, or by default}

+

 \item{zscore}{boolean indicating the transformation of gene expression 

 in object to zscore or not}

+\item{target_cell_type}{alternative cell types (in case of testing classifier)}

+

+\item{pos_parent}{a vector indicating parent classifier prediction}

+

+\item{parent_cell_type}{name of parent cell type}

+

 \item{pred_cells}{a whole prediction for all cells}

 \item{ignore_ambiguous_result}{whether ignore ambigouous result}

@@ -174,6 +202,8 @@ in object to zscore or not}

 \item{clusts}{cluster info}

 \item{most_probable_cell_type}{predicted cell type}

+

+\item{verbose}{logical indicating downloading the file or not}

 }

 \value{

 TRUE if the classifier is valid or the reason why it is not

@@ -200,6 +230,10 @@ the classifier with the new marker genes

 scAnnotatR object with the new marker genes.

+caret trained model

+

+model performance statistics

+

 a list of balanced count matrix

 and corresponding tags of balanced count matrix

@@ -207,8 +241,6 @@ the classification model (caret object)

 row wise center-scaled count matrix

-list of classifiers

-

 filtered matrix

 list of adjusted tag

@@ -226,7 +258,18 @@ simplified prediction

 applicable matrix

 classifier performance

+

+BiocFileCache object

+

+path to the downloaded file in cache

 }

 \description{

 Check if a scAnnotatR object is valid

+

+Train a classifier for a new cell type from expression matrix

+and tag 

+If cell type has a parent, only available for \code{\link{scAnnotatR}}

+object as parent cell classifying model.

+

+Testing process from matrix and tag

 }

new file mode 100644

@@ -0,0 +1,17 @@

+% Generated by roxygen2: do not edit by hand

+% Please edit documentation in R/support.R

+\name{load_models}

+\alias{load_models}

+\title{Load classifiers from databases}

+\usage{

+load_models(path_to_models)

+}

+\arguments{

+\item{path_to_models}{path to databases, or by default}

+}

+\value{

+list of classifiers

+}

+\description{

+Load classifiers from databases

+}

@@ -4,10 +4,10 @@

 \alias{plant_tree}

 \title{Plant tree from list of models}

 \usage{

-plant_tree(models.file.path = "default")

+plant_tree(path_to_models = "default")

 }

 \arguments{

-\item{models.file.path}{list of models. If not provided, 

+\item{path_to_models}{list of models. If not provided, 

 list of default pretrained models in the package will be used.}

 }

 \value{

@@ -4,7 +4,7 @@

 \alias{save_new_model}

 \title{Save a model to the package}

 \usage{

-save_new_model(new_model, include.default = TRUE, path.to.models = tempdir())

+save_new_model(new_model, include.default = TRUE, path_to_models = tempdir())

 }

 \arguments{

 \item{new_model}{new model to be added into the classification tree}

@@ -18,7 +18,7 @@ both of them once. In addition, default pretrained models

 of the package cannot be changed or removed. 

 This can be done with the new trained model list.}

-\item{path.to.models}{path to the folder containing the list of new models.}

+\item{path_to_models}{path to the folder containing the list of new models.}

 }

 \value{

 no return value, but the model is now saved to database

@@ -38,7 +38,7 @@ marker_genes = selected_marker_genes_T, cell_type = "t cells")

 # save the trained classifier to system 

 # test classifier can be used before this step

-save_new_model(new_model = classifier_t, path.to.models = tempdir())

+save_new_model(new_model = classifier_t, path_to_models = tempdir())

 # verify if new model has been saved

 print(names(load(file.path(tempdir(), "new_models.rda"))))

@@ -2,7 +2,7 @@ context("scAnnotatR class functions")

 library(scAnnotatR)

 test_that("Set cell type changes cell type", {

-  data("default_models")

+  default_models <- load_models('default')

   classifier_B <- default_models[['B cells']]

   cell_type(classifier_B) <- "b cells"

@@ -10,7 +10,7 @@ test_that("Set cell type changes cell type", {

 })

 test_that("Set probability threshold changes probability threshold", {

-  data("default_models")

+  default_models <- load_models('default')

   classifier_B <- default_models[['B cells']]

   p_thres(classifier_B) <- 0.6

@@ -18,7 +18,7 @@ test_that("Set probability threshold changes probability threshold", {

 })

 test_that("Set classifier changes classifier and marker genes", {

-  data("default_models")

+  default_models <- load_models('default')

   classifier_B <- default_models[['B cells']]

   classifier_T <- default_models[['T cells']]

@@ -64,7 +64,7 @@ library(scAnnotatR)

 The models are stored in the `default_models` object:

 ```{r}

-data("default_models")

+default_models <- load_models("default")

 names(default_models)

 ```

@@ -233,7 +233,7 @@ New classification models can be stored using the `save_new_model` function:

 ```{r}

 # no copy of pretrained models is performed

-save_new_model(new_model = classifier_B, path.to.models = tempdir(),

+save_new_model(new_model = classifier_B, path_to_models = tempdir(),

                include.default = FALSE) 

 ```

@@ -241,7 +241,7 @@ Parameters:

   * **new_model**: The new model that should be added to the database in the

                    specified directory.

-  * **path.to.models**: The directory where the new models should be stored.

+  * **path_to_models**: The directory where the new models should be stored.

   * **include.default**: If set, the default models shipped with the package

                          are added to the database.

@@ -253,7 +253,7 @@ Models can be deleted from the model database using the `delete_model` function:

 ```{r}

 # delete the "B cells" model from the new database

-delete_model("B cells", path.to.models = tempdir())

+delete_model("B cells", path_to_models = tempdir())

 ```

 ## Session Info

@@ -60,7 +60,7 @@ library(scAnnotatR)

 ```

 ```{r}

-data("default_models")

+default_models <- load_models('default')

 classifier_B <- default_models[['B cells']]

 classifier_B

 ```

@@ -171,7 +171,7 @@ parent classifier to the train method:

 for example: *parent_cell = 'B cells'*

   * Users can give name of a model among models available in users' database 

-AND the path to that database, for example: `parent_cell = 'B cells', path.to.models = '.'`

+AND the path to that database, for example: `parent_cell = 'B cells', path_to_models = '.'`

 Train the child classifier:

 ```{r}

@@ -230,15 +230,15 @@ names(default_models)