Removes tidyverse dependencies; adopts SeuratObject package for acces…

…sion

DESCRIPTION

@@ -1,7 +1,7 @@
 Package: scPred
 Type: Package
 Title: scPred - accurate supervised method for cell-type classification from single-cell RNA-seq data
-Version: 1.9.0
+Version: 1.9.1
 Author: c(person("Jose", "Alquicira-Hernandez", email = "[email protected]", rol = "aut, cre", comment = c(ORCID = "0000-0002-9049-7780")))
 Maintainer: Jose Alquicira-Hernandez <[email protected]>
 Description: Provides a series of functions to train a prediction model based on gene expression data to classify single cells.
@@ -11,19 +11,15 @@ Encoding: UTF-8
 LazyData: true
 Imports:
     harmony,
-    Seurat,
+    SeuratObject,
     methods,
     stats,
     ggbeeswarm,
     MLmetrics,
-    dplyr,
-    tidyr,
-    magrittr,
     caret,
     kernlab,
     pROC,
     ggplot2,
-    tibble,
     knitr,
     pbapply,
     cli,

NAMESPACE

@@ -15,30 +15,21 @@ exportMethods(get_metadata)
 exportMethods(get_probabilities)
 exportMethods(get_scpred)
 exportMethods(show)
-importFrom(Seurat,AddMetaData)
-importFrom(Seurat,CreateDimReducObject)
-importFrom(Seurat,DefaultAssay)
-importFrom(Seurat,Embeddings)
-importFrom(Seurat,GetAssayData)
-importFrom(Seurat,Loadings)
-importFrom(Seurat,ScaleData)
-importFrom(Seurat,Stdev)
-importFrom(Seurat,VariableFeatures)
+importFrom(SeuratObject,AddMetaData)
+importFrom(SeuratObject,Cells)
+importFrom(SeuratObject,CreateDimReducObject)
+importFrom(SeuratObject,DefaultAssay)
+importFrom(SeuratObject,Embeddings)
+importFrom(SeuratObject,GetAssayData)
+importFrom(SeuratObject,Loadings)
+importFrom(SeuratObject,Reductions)
+importFrom(SeuratObject,Stdev)
+importFrom(SeuratObject,VariableFeatures)
 importFrom(caret,prSummary)
 importFrom(caret,train)
 importFrom(caret,trainControl)
 importFrom(caret,twoClassSummary)
-importFrom(dplyr,arrange)
-importFrom(dplyr,bind_rows)
 importFrom(dplyr,distinct)
-importFrom(dplyr,filter)
-importFrom(dplyr,full_join)
-importFrom(dplyr,group_by_)
-importFrom(dplyr,if_else)
-importFrom(dplyr,mutate)
-importFrom(dplyr,mutate_all)
-importFrom(dplyr,select)
-importFrom(dplyr,summarise)
 importFrom(ggbeeswarm,geom_quasirandom)
 importFrom(ggplot2,aes)
 importFrom(ggplot2,element_blank)
@@ -53,14 +44,8 @@ importFrom(ggplot2,xlab)
 importFrom(ggplot2,ylab)
 importFrom(harmony,HarmonyMatrix)
 importFrom(knitr,kable)
-importFrom(magrittr,"%>%")
 importFrom(methods,is)
 importFrom(methods,setMethod)
 importFrom(methods,show)
 importFrom(pbapply,pblapply)
 importFrom(stats,predict)
-importFrom(tibble,column_to_rownames)
-importFrom(tibble,rownames_to_column)
-importFrom(tidyr,gather)
-importFrom(tidyr,pivot_longer)
-importFrom(tidyr,spread)

R/getFeatureSpace.R

@@ -18,10 +18,8 @@
 #' }
 #' @keywords informative, significant, features
 #' @importFrom methods is
-#' @importFrom tidyr gather
-#' @importFrom magrittr "%>%"
-#' @importFrom dplyr mutate arrange filter distinct
 #' @importFrom pbapply pblapply
+#' @importFrom SeuratObject DefaultAssay Embeddings Loadings Reductions Cells
 #' 
 #' @export
 #' @author
@@ -49,7 +47,7 @@ getFeatureSpace <- function(object, pvar, correction = "fdr", sig = 1, reduction
     stop("Invalid multiple testing correction method. See ?p.adjust function")
   }
 
-  if(!pvar %in% names(object@meta.data)){
+  if(!pvar %in% names(object[[]])){
     stop("Prediction variable is not stored in metadata slot")
   }
 
@@ -141,8 +139,8 @@ getFeatureSpace <- function(object, pvar, correction = "fdr", sig = 1, reduction
   }else{
 
     res <- pblapply(levels(classes), .getFeatures, classes, cellEmbeddings, correction, sig)
-    dict <- data.frame(classes, original_classes) %>% 
-      distinct()
+    dict <- data.frame(classes, original_classes)
+    dict <- unique(dict)
 
     i <- match(levels(classes), dict$classes)
     names(res) <- as.character(dict$original_classes[i])
@@ -162,10 +160,7 @@ getFeatureSpace <- function(object, pvar, correction = "fdr", sig = 1, reduction
 
   }
 
-
-
   # Create scPred object
-
   spmodel@pvar <- pvar
   spmodel@features <- res
   spmodel@cell_embeddings <- cellEmbeddings
@@ -193,16 +188,15 @@ getFeatureSpace <- function(object, pvar, correction = "fdr", sig = 1, reduction
   # Get indices for positive and negative class cells
   positiveCells <- newClasses == positiveClass
   negativeCells <- newClasses == "other"
+
 
   # Get informative features
-  apply(cellEmbeddings, 2, function(d) stats::wilcox.test(d[positiveCells], d[negativeCells])) %>%
-    lapply('[[', "p.value") %>% # Extract p-values
-    as.data.frame() %>% 
-    gather(key = "feature", value = "pValue") %>%
-    mutate(pValueAdj = stats::p.adjust(pValue, method = correction, n = nrow(.))) %>% # Perform multiple test correction
-    arrange(pValueAdj) %>% 
-    filter(pValueAdj < sig) -> sigDims
-
-  sigDims
+  wt_res <- apply(cellEmbeddings, 2, function(d) stats::wilcox.test(d[positiveCells], d[negativeCells]))
+  wt_res <- lapply(wt_res, '[[', "p.value")
+  wt_res <- data.frame(feature = names(wt_res), pValue = as.numeric(wt_res))
+  wt_res$pValueAdj <- stats::p.adjust(wt_res$pValue, method = correction, n = nrow(wt_res))
+  wt_res <- wt_res[order(wt_res$pValueAdj), ]
+  wt_res <- wt_res[wt_res$pValueAdj < sig, ]
+  wt_res
 }