Merge pull request #352 from biomage-org/fix-manual-spline

Fix number of genes vs UMIs manual spline

pipeline-runner/R/qc-4-filter_gene_umi_outlier.R

@@ -52,7 +52,7 @@ filter_gene_umi_outlier <- function(scdata_list, config, sample_id, cells_id, ta
 
   # get p_level and update in config
   # defaults from "gene.vs.molecule.cell.filter" in pagoda2
-  if (safeTRUE(config$auto))
+  if (safeTRUE(config$auto) || is.null(config$filterSettings$predictionInterval))
     p_level <- min(0.001, 1 / ncol(sample_data))
   else
     p_level <- 1 - as.numeric(config$filterSettings$predictionInterval)

pipeline-runner/tests/testthat/test-qc-4-filter_gene_umi_outlier.R

@@ -184,3 +184,25 @@ test_that("Gene UMI filter throws error if input is a non float-interpretable st
 
   expect_error(filter_gene_umi_outlier(scdata, config, sample_2_id, cells_id))
 })
+
+test_that("Gene UMI filter works with manual settings and default prediction interval value", {
+  c(scdata_list, sample_1_id, sample_2_id) %<-% mock_scdata()
+  config <- mock_config()
+  cells_id <- mock_ids()
+  type <- "spline"
+  config$filterSettings$regressionType <- type
+
+  config$auto <- TRUE
+  out_auto <- filter_gene_umi_outlier(scdata_list, config, sample_1_id, cells_id)
+
+  config$auto <- FALSE
+  out_manual <- filter_gene_umi_outlier(scdata_list, config, sample_1_id, cells_id)
+
+  expect_null(config$filterSettings$predictionInterval)
+  expect_equal(
+    out_auto$config$filterSettings$regressionTypeSettings[[type]]$p.level,
+    out_manual$config$filterSettings$regressionTypeSettings[[type]]$p.level
+  )
+})
+
+