Seurat 4.4

.github/workflows/pr.yaml

@@ -9,7 +9,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        os: ["ubuntu-latest", "macos-latest"]
+        os: ["ubuntu-latest"]
     defaults:
       run:
         shell: bash -l {0}
@@ -20,32 +20,19 @@ jobs:
         run: |
           echo "${GITHUB_WORKSPACE}" >> $GITHUB_PATH
 
-      - name: Cache conda
-        uses: actions/cache@v1
-        env:
-          # Increase this value to reset cache if etc/example-environment.yml has not changed
-          CACHE_NUMBER: 0
-        with:
-          path: ~/conda_pkgs_dir
-          key:
-            ${{ matrix.os }}-conda-${{ env.CACHE_NUMBER }}-${{hashFiles('test-environment.yml') }}
-
-      - uses: conda-incubator/setup-miniconda@v2
+      - uses: mamba-org/setup-micromamba@v1
         with:
-          activate-environment: test
+          micromamba-version: '1.4.5-0'
           environment-file: test-environment.yml
-          mamba-version: "*"
-          miniforge-variant: Mambaforge
-          use-mamba: true
-          miniforge-version: latest
-          python-version: 3.6
-          channels: conda-forge,bioconda,defaults
-          allow-softlinks: true
-          channel-priority: flexible
-          show-channel-urls: true
-          use-only-tar-bz2: true
+          init-shell: >-
+            bash
+          cache-environment: true
+          create-args: |
+            python=${{ matrix.python-version }}
+          post-cleanup: 'all'
 
       - name: Run tests
+        shell: bash -el {0}
         env:
           GITHUB_PAT: ${{ secrets.GITHUB_TOKEN }}
         run: |

README.md

@@ -1,12 +1,14 @@
-# seurat-scripts 4.x.y for Seurat 4.0.0 [![install with bioconda](https://img.shields.io/badge/install%20with-bioconda-brightgreen.svg?style=flat)](http://bioconda.github.io/recipes/seurat-scripts/README.html)
+# seurat-scripts 4.x.y for Seurat 4.4.0 [![install with bioconda](https://img.shields.io/badge/install%20with-bioconda-brightgreen.svg?style=flat)](http://bioconda.github.io/recipes/seurat-scripts/README.html)
 
 In order to wrap Seurat's internal workflow in any given workflow language, it's important to have scripts to call each of those steps, which is what this package provides.
 
 This version of seurat-scripts uses native conversions to Loom, SCE, Seurat H5 and AnnData.
 
+Please note that given the little man power available to maintain this, we dropped MacOS support for Seurat version 4.4.0 onwards (not all tests were passing and there was no bandwidth to take care of the dependencies issues on both linux and macOS).
+
 ## Install
 
-The recommended method for script installation is via a Bioconda recipe called seurat-scripts.
+The recommended method for usage is through the [biocontainer](https://quay.io/repository/biocontainers/seurat-scripts?tab=tags&tag=latest) produced by Bioconda. If needed, scripts installation is via a Bioconda recipe called seurat-scripts.
 
 With the [Bioconda channels](https://bioconda.github.io/#set-up-channels) configured the latest release version of the package can be installed via the regular conda install command:
 

r-seurat-scripts-post-install-tests.bats

@@ -309,6 +309,7 @@
 }
 
 @test "Classify against reference" {
+     # maybe this should be moved after PCA and UMAP of integrated object
      if [ "$use_existing_outputs" = 'true' ] && [ -f "$classify_result_object" ]; then
          skip "$classify_result_object exists and use_existing_outputs is set to true"
      fi
@@ -466,7 +467,7 @@
       -o $tmp_conserved_markers_fn_object && \
     seurat-find-clusters.R -i $tmp_conserved_markers_fn_object --resolution 0.5 \
       -o $tmp_conserved_markers_cl_object -t $cluster_text_file".conserved_tmp.txt" && \
-    seurat-find-conserved-markers.R -i $tmp_conserved_markers_cl_object --ident-1 6 --ident-2 4 --grouping-var "tech" -o $conserved_markers_result
+    seurat-find-conserved-markers.R -i $tmp_conserved_markers_cl_object --ident-1 1 --ident-2 2 --grouping-var "tech" -o $conserved_markers_result
 
   [ "$status" -eq 0 ]
   [ -f "$conserved_markers_result" ]
@@ -515,7 +516,7 @@
     run rm -rf ${loom_converted_cluster_object}.loom && \
       seurat-convert.R -i $singlecellexperiment_converted_cluster_object \
         --input-format singlecellexperiment -o $loom_converted_cluster_object \
-        --output-format loom
+        --output-format loom --input-assay NULL
 
     echo "status = ${status}"
     echo "output = ${output}"

seurat-classify-against-reference.R

@@ -93,6 +93,13 @@ option_list = list(
     type = 'character',
     help = "Dimensional reduction to perform when finding anchors."
   ),
+  make_option(
+    c("--reference_reduction"),
+    action = "store",
+    default = "pca",
+    type = 'character',
+    help = "Name of dimensional reduction to use from the reference if running the pcaproject workflow. Optionally enables reuse of precomputed reference dimensional reduction. If NULL (default), use a PCA computed on the reference object."
+  ),
   make_option(
     c("--project-query"),
     action = "store_false",
@@ -285,13 +292,24 @@ if (opt$query_format == "singlecellexperiment" | opt$reference_format == "single
 
 seurat_query <- read_seurat4_object(input_path = opt$query_file, format = opt$query_format)
 seurat_reference <- read_seurat4_object(input_path = opt$reference_file, format = opt$reference_format)
+
+if(!(opt$reference_reduction %in% names(seurat_reference@reductions))) {
+  print(paste0("Calculating reduction ", opt$reference_reduction," as it is not present in the reference object."))
+  if(opt$reference_reduction == "pca") {
+    seurat_reference <- ScaleData(seurat_reference, verbose = FALSE)
+    seurat_reference <- RunPCA(seurat_reference, npcs=30, verbose = FALSE)
+  } else {
+    stop((paste('Reduction default calculation on reference not implemented for ', opt$reference_reduction, ', please compute previously on reference object.')))
+  }
+}
 #make the function work
 anchor_object <- FindTransferAnchors(seurat_reference,
                                     seurat_query,
                                     normalization.method = opt$normalization_method,
                                     reference.assay = opt$reference_assay,
                                     query.assay = opt$query_assay,
                                     reduction = opt$reduction,
+                                    reference.reduction = opt$reference_reduction,
                                     project.query = opt$project_query,
                                     features = opt$features,
                                     npcs = opt$npcs,

seurat-convert.R

@@ -39,6 +39,13 @@ option_list <- list(
                     default = "seurat",
                     type = "character",
                     help = "Either seurat, h5seurat, loom, singlecellexperiment or h5seurat (partial support)")
+,
+    make_option(
+                    c("--input-assay"),
+                    action = "store",
+                    default = "RNA",
+                    type = "character",
+                    help = "The assay to be read from the input object, if it corresponds")
 )
 
 opt <- wsc_parse_args(option_list,
@@ -49,9 +56,12 @@ if (!file.exists(opt$input_object_file)) {
     stop((paste("File", opt$input_object_file, "does not exist")))
 }
 
+if(opt$input_assay == "NULL") {
+    opt$input_assay <- NULL
+}
 
 seurat_object <- read_seurat4_object(input_path = opt$input_object_file,
-                    format = opt$input_format)
+                    format = opt$input_format, assay = opt$input_assay)
 
 write_seurat4_object(seurat_object = seurat_object,
                     output_path = opt$output_object_file,

seurat-find-conserved-markers.R

@@ -10,6 +10,7 @@ suppressPackageStartupMessages(require(optparse))
 suppressPackageStartupMessages(require(scater))
 suppressPackageStartupMessages(require(Seurat))
 suppressPackageStartupMessages(require(workflowscriptscommon))
+suppressPackageStartupMessages(require(metap))
 
 option_list <- list(
     make_option(
@@ -65,9 +66,9 @@ option_list <- list(
     make_option(
         c("--meta-method"),
         action = "store",
-        default = "metap::minimump",
+        default = "minimump",
         type = "character",
-        help = "method for combining p-values. Should be a function from the metap package (NOTE: pass the function, not a string)"
+        help = "method for combining p-values. Should be a function from the metap package: invchisq, invt, logitp, meanp, meanz, maximump, minimump, sumlog, sumz, truncated, votep, wilkinsonp"
     ),
     make_option(
         c("--reduction"),
@@ -204,13 +205,19 @@ if (!file.exists(opt$input_object_file)) {
     stop((paste("File", opt$input_object_file, "does not exist")))
 }
 
+list_of_meta_methods <- ls("package:metap")
+metap_method <- NULL
+if(!is.null(opt$meta_method) && !(opt$meta_method %in% list_of_meta_methods)) {
+    stop((paste("Metap method requested ", opt$meta_method, " doesn't exist, please choose from: ", list_of_meta_methods)))
+} else {
+    # safer way of running eval parse
+    opt$meta_method <- eval(parse(text=paste0("metap::", opt$meta_method)))
+}
 
 load_seurat4_packages_for_format(formats = c(opt$query_format, opt$anchors_format, opt$reference_format))
 
 seurat_object <- read_seurat4_object(input_path = opt$input_object_file,
                     format = opt$input_format)
-# transform method name into actual R function. This should probably be sanitised in Galaxy
-opt$meta_method <- eval(parse(text=opt$meta_method))
 
 conserved_markers <- FindConservedMarkers(object = seurat_object,
                     ident.1 = opt$ident_1,

test-environment.yml

@@ -1,17 +1,22 @@
-name: test
+name: [email protected]
+channels:
+  - defaults
+  - conda-forge
+  - bioconda
 dependencies:
-  - bats
+  - bats-core
   - r-optparse
   - libpng
   - r-cairo
   - r-workflowscriptscommon >=0.0.8
-  - r-seurat=4.0.0
-  - r-uwot <0.1.11
+  - r-seurat<=4.4
+  # - r-uwot
   - r-loom
   - r-svglite
   - bioconductor-scater
   - r-seuratdisk
   - r-remotes
   - bioconductor-multtest
   - r-metap
-  - pandoc
+  - pandoc
+  - bioconductor-limma