Merge pull request #3 from neurogenomics/nfancy-patch-1

Nfancy patch 1

.github/workflows/r_package_dev-nf.yaml → .github/workflows/workflow_master.yaml

@@ -3,7 +3,8 @@ name: R GitHub Actions dev-nf
 on:
   push:
     branches:
-      - dev-nf
+      - master
+      - nfancy-patch-1
 
 jobs:
   R-CMD-check:
@@ -35,10 +36,7 @@ jobs:
           load: true
           file: ./Dockerfile
           tags: |
-            nfancy/scflow:${{ env.packageVersion }}
             nfancy/scflow:latest
-            nfancy/scflow:dev
-
 
 
       - name: Login to DockerHub
@@ -49,6 +47,4 @@ jobs:
 
       - name: push service
         run: |
-          docker push nfancy/scflow:${{ env.packageVersion }}
           docker push nfancy/scflow:latest
-          docker push nfancy/scflow:dev

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: scFlow
 Title: A Single-Cell/Nuclei Analysis Toolkit
-Version: 0.7.2
+Version: 0.7.4
 Authors@R: 
     c(
     person(given = "Combiz",
@@ -41,21 +41,21 @@ Suggests:
     spelling,
     covr,
     knitr,
-    scFlowExamples,
     BiocStyle,
     DT (>= 0.26)
 Language: en-US
-RoxygenNote: 7.2.3
+RoxygenNote: 7.3.1
 biocViews:
 Imports: 
+    Matrix,
     assertthat (>= 0.2.1),
     bib2df,
     biomaRt (>= 2.54.0),
     cli (>= 3.4.1),
     cowplot (>= 1.1.1),
     DirichletReg (>= 0.7.1),
-    DoubletFinder (>= 2.0.3),
     dplyr (>= 1.1.0),
+    DoubletFinder (>= 2.0.3),
     DropletUtils (>= 1.18.1),
     edgeR (>= 3.40.0),
     english (>= 1.2.6),
@@ -75,7 +75,7 @@ Imports:
     limma (>= 3.54.0),
     magrittr (>= 2.0.3),
     MAST (>= 1.24.0),
-    Matrix,
+    lme4,
     methods,
     monocle3 (>= 1.3.1),
     paletteer (>= 1.5.0),
@@ -106,16 +106,11 @@ Imports:
     uwot,
     WebGestaltR
 Remotes:
+    NathanSkene/EWCE,
     chris-mcginnis-ucsf/DoubletFinder,
     ropensci/plotly,
     cole-trapnell-lab/monocle3,
     theislab/kBET,
     jlmelville/uwot,
-    NathanSkene/One2One,
-    hhoeflin/hdf5r,
-    mojaveazure/loomR,
-    neurogenomics/scFlowExamples,
-    ropensci/bib2df,
-    bzhanglab/WebGestaltR,
-    cran/riverplot
+    ropensci/bib2df
 VignetteBuilder: knitr

Dockerfile

@@ -1,8 +1,8 @@
 #LABEL maintainer="Combiz Khozoie, Ph.D. [email protected], Alan Murphy, [email protected]"
 
 ## Use rstudio installs binaries from RStudio's RSPM service by default,
-## Uses the latest stable ubuntu, R and Bioconductor versions. Created on unbuntu 20.04, R 4.0 and BiocManager 3.12
-FROM rocker/rstudio:4.2.2
+## Uses the latest stable ubuntu, R and Bioconductor versions. Created on unbuntu 20.04, R 4.3 and BiocManager 3.18
+FROM rocker/rstudio:4.3
 
 
 ## Add packages dependencies
@@ -38,7 +38,7 @@ RUN apt-get update \
 	libcairo2-dev \
 	libtiff5-dev \
 	libreadline-dev \
-	libgsl0-dev \
+	libgsl-dev \
 	libgslcblas0 \
 	libgtk2.0-dev \
 	libgl1-mesa-dev \
@@ -111,7 +111,8 @@ RUN pip install stratocumulus \
 && rm -rf /tmp/*
 
 
-RUN install2.r -e \
+RUN install2.r -e -t source \
+Matrix \
 argparse \
 assertthat \
 BiocManager \
@@ -138,12 +139,11 @@ ggridges \
 Hmisc \
 httr \
 ids \
-igraph \
 knitr \
 leaflet \
-lme4 \
 magrittr \
-Matrix \
+lme4 \
+igraph \
 paletteer \
 patchwork \
 plyr \
@@ -166,6 +166,7 @@ Rtsne \
 scales \
 sctransform \
 Seurat \
+SeuratObject \
 snow \
 spelling \
 stringr \
@@ -178,6 +179,8 @@ tidyverse \
 UpSetR \
 utils \
 vroom \
+WebGestaltR \
+apcluster \
 && rm -rf /tmp/downloaded_packages
 
 ## Install Bioconductor packages
@@ -187,18 +190,15 @@ RUN Rscript -e 'requireNamespace("BiocManager"); BiocManager::install(ask=F);' \
 && rm -rf /tmp/downloaded_packages
 
 ## Install from GH the following
-RUN installGithub.r chris-mcginnis-ucsf/DoubletFinder \
+RUN installGithub.r NathanSkene/EWCE \
+chris-mcginnis-ucsf/DoubletFinder \
 ropensci/plotly \
 cole-trapnell-lab/monocle3 \
 theislab/kBET \
 jlmelville/uwot \
-NathanSkene/One2One \
 hhoeflin/hdf5r \
-mojaveazure/loomR \
 ropensci/bib2df \
 cvarrichio/Matrix.utils \
-neurogenomics/scFlowExamples \
-bzhanglab/WebGestaltR \
 && rm -rf /tmp/downloaded_packages
 
 ## Install scFlow package

NEWS.md

@@ -1,3 +1,9 @@
+# scFlow 0.7.4
+
+* Updated DoubletFinder version compatibility.
+* Minor bug fixes in `map_celltypes_sce()`. The subsampling is now done per cluster to avoid smaller clusters being excluded.
+* Change the order of the contrast vs reference names in DEG table.
+
 # scFlow 0.7.2
 
 * QC bug fixes.

R/cluster_sce.R

@@ -2,8 +2,16 @@
 #' Cluster SingleCellExperiment with monocle3::cluster_cells
 #'
 #' @param sce a SingleCellExperiment object
-#' @param ... see uwot::umap for umap options. Includes reduction_methods one
-#' or more of "PCA", "tSNE", "UMAP", "UMAP3D"
+#' @param cluster_method Available methods include leiden and louvain.
+#' @param reduction_method Input slot for clustering. Available options are PCA,
+#' UMAP_Liger, tSNE_Liger.
+#' @param resolution Clustering resolution. If NULL, clustering method will be
+#' set to louvain.
+#' @param K Integer number of nearest neighbors to use when creating the k
+#' nearest neighbor graph for Louvain/Leiden clustering.
+#' k is related to the resolution of the clustering result, a bigger k will
+#' result in lower resolution and vice versa. Default is 50.
+#' @param ... see monocle3::cluster_cells for more clustering options.
 #'
 #' @return sce a SingleCellExperiment object annotated with reducedDims
 #'
@@ -14,15 +22,22 @@
 #'
 #' @export
 
-cluster_sce <- function(sce, ...) {
+cluster_sce <- function(sce,
+                        cluster_method = "leiden",
+                        reduction_method = "UMAP_Liger",
+                        resolution = 1e-5,
+                        k = 50,
+                        louvain_iter = 1,
+                        verbose = T,
+                        ...) {
 
   fargs <- list(
-    cluster_method = "leiden",
-    reduction_method = "UMAP_Liger",
-    resolution = 1e-5,
-    k = 100,
-    louvain_iter = 1,
-    verbose = T
+    cluster_method = cluster_method,
+    reduction_method = reduction_method,
+    resolution = resolution,
+    k = k,
+    louvain_iter = louvain_iter,
+    verbose = verbose
   )
 
   inargs <- list(...)

R/diffexp_models.R

@@ -413,7 +413,7 @@ perform_de <- function(sce,
       by = "ensembl_gene_id")
     }
 
-    element_name <- paste(fargs$ref_class, ctrast, sep = "_vs_")
+    element_name <- paste(ctrast, fargs$ref_class, sep = "_vs_")
 
     de_params <- list(
       celltype = unique(fargs$sce$cluster_celltype),

R/integrate_sce.R

@@ -4,9 +4,9 @@
 #' Produces a reducedDim slot representing common factors across datasets
 #'
 #' @param sce a SingleCellExperiment object or merged sce objects
-#' @param method the integration method to use
-#'   Set to k=30 as default.
-#' @param ... Additional arguments. Such as k Inner dimension of factorization 
+#' @param method the integration method to use. Set to liger as default.
+#' See `liger_preprocess` and `liger_reduce_dims` functions for parameter options.
+#' @param ... Additional arguments. Such as k, inner dimension of factorization
 #' (number of factors).
 #'
 #' @return sce SingleCellExperiment object annotated with reducedDims

R/map_with_ewce.R

@@ -94,10 +94,20 @@ map_celltypes_sce <- function(sce,
 
   l_ctd <- .read_rds_files_to_list(ctd_folder)
 
-  if (dim(sce)[[2]] > cells_to_sample) {
+
+  if(!is.null(cells_to_sample)){
+  set.seed(123)
+  idx <- c()
+  for(i in as.character(unique(sce[[clusters_colname]]))){
+    idx_temp <- which( sce[[clusters_colname]] == i) %>%
+      sample(size = min(cells_to_sample, length(.)), replace = FALSE )
+    idx_temp <- idx_temp[order(idx_temp)]
+    idx <- c(idx, idx_temp)}
+
     message(sprintf("Subsetting %s cells", cells_to_sample))
-    set.seed(42)
-    sce_subset <- sce[, sample(dim(sce)[[2]], cells_to_sample)]
+
+    sce_subset <- sce[, idx]
+
   } else {
     sce_subset <- sce
   }

README.Rmd

@@ -90,6 +90,23 @@ You can install the development version of scFlow from GitHub with: -
 devtools::install_github("neurogenomics/scFlow")
 ```
 
+## Running RStudio Server with pre-installed scFlow via Docker 
+
+If using linux system, open a cmd terminal and run the following commands:
+
+```{r eval = FALSE}
+docker pull nfancy/scflow
+
+docker run --rm -d \
+-v $HOME:/home/rstudio/home \
+-e ROOT=true \
+-e PASSWORD=password \
+-p 8950:8787 \
+nfancy/scflow
+```
+
+Then access the RStudio Server on your browser at IP:8950. The default username is `rstudio` and password is `password`. To understand more on how to run docker Rstudio follow the link [here](https://davetang.org/muse/2021/04/24/running-rstudio-server-with-docker/). 
+
 ## Support
 
 * Ask a question on Stack Overflow with the scFlow tag, we monitor this for new questions.

README.md

@@ -115,6 +115,27 @@ You can install the development version of scFlow from GitHub with: -
 devtools::install_github("neurogenomics/scFlow")
 ```
 
+## Running RStudio Server with pre-installed scFlow via Docker
+
+If using linux system, open a cmd terminal and run the following
+commands:
+
+``` r
+docker pull nfancy/scflow
+
+docker run --rm -d \
+-v $HOME:/home/rstudio/home \
+-e ROOT=true \
+-e PASSWORD=password \
+-p 8950:8787 \
+nfancy/scflow
+```
+
+Then access the RStudio Server on your browser at IP:8950. The default
+username is `rstudio` and password is `password`. To understand more on
+how to run docker Rstudio follow the link
+[here](https://davetang.org/muse/2021/04/24/running-rstudio-server-with-docker/).
+
 ## Support
 
 -   Ask a question on Stack Overflow with the scFlow tag, we monitor