Merge pull request #177 from cole-trapnell-lab/develop

Pull for v0.1.3

.gitignore

@@ -3,3 +3,4 @@ inst/doc
 .Rhistory
 .RData
 .Ruserdata
+.DS_store

.travis.yml

@@ -5,6 +5,7 @@ language: r
 before_install:
   - sudo apt-get install -y libudunits2-dev
   - sudo apt-get install -y gdal-bin
+  - sudo apt-get install -y libgdal1-dev
 r:
 - bioc-devel
 - bioc-release
@@ -44,6 +45,12 @@ r_binary_packages:
 - tibble
 - tidyr
 - viridis
+
+script:
+  - |
+    R CMD build .
+    travis_wait 20 R CMD check monocle3*tar.gz
+
 after_success:
 - Rscript -e 'covr::codecov()'
 notifications:

DESCRIPTION

@@ -1,7 +1,7 @@
 Package: monocle3
 Title: Clustering, differential expression, and trajectory analysis for single-
     cell RNA-Seq
-Version: 0.1.2
+Version: 0.1.3
 Authors@R: 
     person(given = "Hannah",
            family = "Pliner",
@@ -65,6 +65,7 @@ Imports:
     RcppParallel,
     reshape2 (>= 1.4.3),
     reticulate (>= 1.11.1),
+    rsample (>= 0.0.5),
     RhpcBLASctl,
     Rtsne (>= 0.15),
     S4Vectors,
@@ -86,3 +87,5 @@ Suggests:
     spelling
 VignetteBuilder: knitr
 Language: en-US
+Remotes:
+    VPetukhov/ggrastr

NAMESPACE

@@ -7,6 +7,7 @@ export("principal_graph_aux<-")
 export(aggregate_gene_expression)
 export(choose_cells)
 export(choose_graph_segments)
+export(clear_cds_slots)
 export(cluster_cells)
 export(clusters)
 export(coefficient_table)

NEWS.md

@@ -1,3 +1,16 @@
+# monocle3 0.1.3
+
+### Changes
+* Added bootstrap bars to plot_percent_cells_positive().
+* Added a UMI cutoff for load_cellranger_data().
+* Added clear_cds_slots() to clear slots from a cds.
+* Added cell_stroke parameter to plot_cells.
+
+### Bug fixes
+* Fixed #154, which actually resulted from a problem in how ncenter was calculated for each partition.
+* Fixed an issue that prevents plotting cells by pseudotime when root_cells are passed to order_cells().
+* Fixed #167 - NAs from plot_genes_by_group.
+
 # monocle3 0.1.2
 
 ### Changes

R/graph_test.R

@@ -75,12 +75,12 @@ graph_test <- function(cds,
 
     test_res <- tryCatch({
       if(method == "Moran_I") {
-        mt <- my.moran.test(exprs_val, lw, wc, alternative = alternative)
+        mt <- suppressWarnings(my.moran.test(exprs_val, lw, wc, alternative = alternative))
         data.frame(status = 'OK', p_value = mt$p.value,
                    morans_test_statistic = mt$statistic,
                    morans_I = mt$estimate[["Moran I statistic"]])
       } else if(method == 'Geary_C') {
-        gt <- my.geary.test(exprs_val, lw, wc, alternative = alternative)
+        gt <- suppressWarnings(my.geary.test(exprs_val, lw, wc, alternative = alternative))
         data.frame(status = 'OK', p_value = gt$p.value,
                    geary_test_statistic = gt$statistic,
                    geary_C = gt$estimate[["Geary C statistic"]])

R/learn_graph.R

@@ -226,27 +226,27 @@ multi_component_RGE <- function(cds,
       num_clusters_in_partition <-
         length(unique(clusters(cds)[colnames(X_subset)]))
       num_cells_in_partition = ncol(X_subset)
-      ncenter <- cal_ncenter(num_clusters_in_partition, num_cells_in_partition)
-      if(is.null(ncenter) || ncenter >= ncol(X_subset)) {
-        ncenter <- ncol(X_subset) - 1
+      curr_ncenter <- cal_ncenter(num_clusters_in_partition, num_cells_in_partition)
+      if(is.null(curr_ncenter) || curr_ncenter >= ncol(X_subset)) {
+        curr_ncenter <- ncol(X_subset) - 1
       }
     } else {
-      ncenter <- min(ncol(X_subset) - 1, ncenter)
+      curr_ncenter <- min(ncol(X_subset) - 1, ncenter)
     }
     if (verbose)
-      message(paste("Using", ncenter, "nodes for principal graph"))
+      message(paste("Using", curr_ncenter, "nodes for principal graph"))
 
     kmean_res <- NULL
 
-    centers <- t(X_subset)[seq(1, ncol(X_subset), length.out=ncenter), ,
+    centers <- t(X_subset)[seq(1, ncol(X_subset), length.out=curr_ncenter), ,
                            drop = F]
     centers <- centers + matrix(stats::rnorm(length(centers), sd = 1e-10),
                                 nrow = nrow(centers)) # add random noise
 
     kmean_res <- tryCatch({
       stats::kmeans(t(X_subset), centers=centers, iter.max = 100)
     }, error = function(err) {
-      stats::kmeans(t(X_subset), centers = ncenter, iter.max = 100)
+      stats::kmeans(t(X_subset), centers = curr_ncenter, iter.max = 100)
     })
 
     if (kmean_res$ifault != 0){

R/load_cellranger_data.R

@@ -33,6 +33,7 @@ get_genome_in_matrix_path <- function(matrix_path, genome=NULL) {
 #' @param pipestance_path Path to the output directory produced by Cell Ranger
 #' @param genome The desired genome (e.g., 'hg19' or 'mm10')
 #' @param barcode_filtered Load only the cell-containing barcodes
+#' @param umi_cutoff Numeric, desired cutoff to include a cell. Default is 100.
 #' @return a new cell_data_set object
 #' @export
 #' @examples
@@ -41,7 +42,7 @@ get_genome_in_matrix_path <- function(matrix_path, genome=NULL) {
 #' gene_bc_matrix <- load_cellranger_data("/home/user/cellranger_output")
 #' }
 load_cellranger_data <- function(pipestance_path=NULL, genome=NULL,
-                                 barcode_filtered=TRUE) {
+                                 barcode_filtered=TRUE, umi_cutoff = 100) {
   # check for correct directory structure
   if (!dir.exists(pipestance_path))
     stop("Could not find the pipestance path: '", pipestance_path,"'.
@@ -132,8 +133,11 @@ load_cellranger_data <- function(pipestance_path=NULL, genome=NULL,
   barcodes$V1 = make.unique(barcodes$V1)
   colnames(data) = barcodes[,1]
   pd = data.frame(barcode=barcodes[,1], row.names=barcodes[,1])
+  data <- data[,Matrix::colSums(data) > umi_cutoff]
+  pd <- pd[colnames(data),, drop=FALSE]
   gbm <- new_cell_data_set(data,
                         cell_metadata = pd,
                         gene_metadata =  feature.names)
+
   return(gbm)
 }

R/order_cells.R

@@ -93,7 +93,8 @@ order_cells <- function(cds,
   } else if(!is.null(root_cells)){
     closest_vertex <- cds@principal_graph_aux[[
       reduction_method]]$pr_graph_cell_proj_closest_vertex
-    root_pr_nodes <- paste("Y_", closest_vertex[root_cells,], sep="")
+    root_pr_nodes <- unique(paste("Y_", closest_vertex[root_cells,], sep=""))
+    #principal_graph(cds)[[reduction_method]]
   }
 
   cds@principal_graph_aux[[reduction_method]]$root_pr_nodes <- root_pr_nodes