Merge pull request #66 from biomage-org/subset-experiment-flag

[BIOMAGE-2253] Subset experiment flag

pipeline-runner/R/gem2s-5-create_seurat.R

@@ -38,6 +38,7 @@ create_seurat <- function(input, pipeline_config, prev_out) {
 
 
   prev_out$scdata_list <- scdata_list
+  prev_out$disable_qc_filters <- FALSE
   res <- list(
     data = list(),
     output = prev_out

pipeline-runner/R/gem2s-6-construct_qc_config.R

@@ -6,15 +6,17 @@
 #'
 #' @param scdata_list list of seurat objects
 #' @param any_filtered bool indicating if barcodes were filtered by emptyDrops
+#' @param disable_qc_filters bool indicating if the data derives from the
+#' subsetting of another experiment
 #'
 #' @return list of QC configuration parameters
 #'
-construct_qc_config <- function(scdata_list, any_filtered) {
+construct_qc_config <- function(scdata_list, any_filtered, disable_qc_filters) {
   samples <- names(scdata_list)
 
   # classifier
   classifier_config_to_duplicate <- list(
-    enabled = !any_filtered,
+    enabled = !any_filtered && !disable_qc_filters,
     prefiltered = any_filtered,
     auto = TRUE,
     filterSettings = list(FDR = 0.01)
@@ -34,7 +36,7 @@ construct_qc_config <- function(scdata_list, any_filtered) {
 
   # mito
   default_mitochondrialContent_config <- list(
-    enabled = TRUE,
+    enabled = !disable_qc_filters,
     auto = TRUE,
     filterSettings = list(
       method = "absoluteThreshold",
@@ -51,7 +53,7 @@ construct_qc_config <- function(scdata_list, any_filtered) {
 
   # ngenes vs umis
   default_numGenesVsNumUmis_config <- list(
-    enabled = TRUE,
+    enabled = !disable_qc_filters,
     auto = TRUE,
     filterSettings = list(
       regressionType = "linear",
@@ -67,7 +69,7 @@ construct_qc_config <- function(scdata_list, any_filtered) {
 
   # doublet scores
   default_doubletScores_config <- list(
-    enabled = TRUE,
+    enabled = !disable_qc_filters,
     auto = TRUE,
     filterSettings = list(
       probabilityThreshold = 0.5,
@@ -140,7 +142,7 @@ get_cellsize_config <- function(scdata_list, config) {
 
 get_sample_mitochondrial_config <- function(scdata_list.sample, config) {
   config.sample <- list(
-    enabled = TRUE,
+    enabled = config$enabled,
     auto = TRUE,
     filterSettings = list(
       method = "absoluteThreshold",

pipeline-runner/R/gem2s-6-prepare_experiment.R

@@ -15,10 +15,11 @@
 prepare_experiment <- function(input, pipeline_config, prev_out) {
   message("Preparing experiment ...")
 
-  check_names <- c("config", "counts_list", "annot", "doublet_scores", "scdata_list")
+  check_names <- c("config", "counts_list", "annot", "doublet_scores", "scdata_list", "disable_qc_filters")
   check_prev_out(prev_out, check_names)
 
   scdata_list <- prev_out$scdata_list
+  disable_qc_filters <- prev_out$disable_qc_filters
   samples <- names(scdata_list)
 
   message("Total cells:", sum(sapply(scdata_list, ncol)))
@@ -29,7 +30,7 @@ prepare_experiment <- function(input, pipeline_config, prev_out) {
   # construct default QC config and update prev out
   message("Constructing default QC configuration...")
   any_filtered <- !(length(prev_out$edrops) == length(samples))
-  prev_out$qc_config <- construct_qc_config(scdata_list, any_filtered)
+  prev_out$qc_config <- construct_qc_config(scdata_list, any_filtered, disable_qc_filters)
 
   res <- list(
     data = list(),

pipeline-runner/R/gem2s-7-upload_to_aws.R

@@ -10,7 +10,7 @@
 #'
 upload_to_aws <- function(input, pipeline_config, prev_out) {
   message("Uploading to AWS ...")
-  check_names <- c("config", "counts_list", "annot", "doublet_scores", "scdata_list", "qc_config")
+  check_names <- c("config", "counts_list", "annot", "doublet_scores", "scdata_list", "qc_config", "disable_qc_filters")
   check_prev_out(prev_out, check_names)
 
   experiment_id <- input$experimentId
@@ -20,6 +20,7 @@ upload_to_aws <- function(input, pipeline_config, prev_out) {
   scdata_list <- prev_out$scdata_list
   config <- prev_out$config
   qc_config <- prev_out$qc_config
+  disable_qc_filters <- prev_out$disable_qc_filters
 
   message("Constructing cell sets ...")
   cell_sets <- get_cell_sets(scdata_list, input)

pipeline-runner/tests/testthat/test-gem2s-6-construct_qc_config.R

@@ -26,7 +26,7 @@ mock_scdata_list <- function() {
 
 test_that("cellsize filter is disabled by default and classifier is pre-filtered", {
   scdata_list <- mock_scdata_list()
-  qc_config <- construct_qc_config(scdata_list, any_filtered = TRUE)
+  qc_config <- construct_qc_config(scdata_list, any_filtered = TRUE, disable_qc_filters = FALSE)
 
   for (sample in names(scdata_list)) {
     expect_false(qc_config$classifier[[sample]]$enabled)
@@ -38,7 +38,7 @@ test_that("cellsize filter is disabled by default and classifier is pre-filtered
 
 test_that("cellsize filter is disabled by default and classifier is not pre-filtered", {
   scdata_list <- mock_scdata_list()
-  qc_config <- construct_qc_config(scdata_list, any_filtered = FALSE)
+  qc_config <- construct_qc_config(scdata_list, any_filtered = FALSE, disable_qc_filters = FALSE)
 
   for (sample in names(scdata_list)) {
     expect_false(qc_config$cellSizeDistribution[[sample]]$enabled)
@@ -47,3 +47,30 @@ test_that("cellsize filter is disabled by default and classifier is not pre-filt
   }
 })
 
+
+test_that("all filters are disabled when disable_qc_filters = TRUE and classifier is pre-filtered", {
+  scdata_list <- mock_scdata_list()
+  qc_config <- construct_qc_config(scdata_list, any_filtered = TRUE, disable_qc_filters = TRUE)
+
+  for (sample in names(scdata_list)) {
+    expect_false(qc_config$cellSizeDistribution[[sample]]$enabled)
+    expect_false(qc_config$mitochondrialContent[[sample]]$enabled)
+    expect_false(qc_config$classifier[[sample]]$enabled)
+    expect_false(qc_config$numGenesVsNumUmis[[sample]]$enabled)
+    expect_false(qc_config$doubletScores[[sample]]$enabled)
+  }
+})
+
+
+test_that("all filters are disabled when disable_qc_filters = TRUE and classifier is not pre-filtered", {
+  scdata_list <- mock_scdata_list()
+  qc_config <- construct_qc_config(scdata_list, any_filtered = FALSE, disable_qc_filters = TRUE)
+
+  for (sample in names(scdata_list)) {
+    expect_false(qc_config$cellSizeDistribution[[sample]]$enabled)
+    expect_false(qc_config$mitochondrialContent[[sample]]$enabled)
+    expect_false(qc_config$classifier[[sample]]$enabled)
+    expect_false(qc_config$numGenesVsNumUmis[[sample]]$enabled)
+    expect_false(qc_config$doubletScores[[sample]]$enabled)
+  }
+})

pipeline-runner/tests/testthat/test-gem2s-7-upload_to_aws.R

@@ -231,7 +231,8 @@ test_that("upload_to_aws tries to upload the correct files to aws", {
                    annot = list(),
                    doublet_scores = list(),
                    scdata_list = scdata_list,
-                   qc_config = list("mock_qc_config"))
+                   qc_config = list("mock_qc_config"),
+                   disable_qc_filters = FALSE)
 
   res <- stubbed_upload_to_aws(input, pipeline_config, prev_out)