Adding annotation file param to bigDIANNtoMSstatsFormat

R/clean_DIANN.R

@@ -7,13 +7,15 @@
 #' @param global_qvalue_cutoff Global Q-value cutoff
 #' @param qvalue_cutoff Q-value cutoff
 #' @param pg_qvalue_cutoff Protein group Q-value cutoff
+#' @param annotation Annotation file or data frame
 #' @return NULL. Writes to file.
 #' @keywords internal
 reduceBigDIANN <- function(input_file, output_path, MBR = TRUE,
                            quantificationColumn = "FragmentQuantCorrected",
                            global_qvalue_cutoff = 0.01,
                            qvalue_cutoff = 0.01,
-                           pg_qvalue_cutoff = 0.01) {
+                           pg_qvalue_cutoff = 0.01,
+                           annotation = NULL) {
   if (grepl("csv", input_file)) {
     delim = ","
   } else if (grepl("tsv|xls", input_file)) {
@@ -23,7 +25,7 @@ reduceBigDIANN <- function(input_file, output_path, MBR = TRUE,
   }
 
   diann_chunk <- function(x, pos) cleanDIANNChunk(x, output_path, MBR, quantificationColumn, pos,
-                     global_qvalue_cutoff, qvalue_cutoff, pg_qvalue_cutoff)
+                     global_qvalue_cutoff, qvalue_cutoff, pg_qvalue_cutoff, annotation)
 
   readr::read_delim_chunked(input_file,
                             readr::DataFrameCallback$new(diann_chunk),
@@ -41,13 +43,15 @@ reduceBigDIANN <- function(input_file, output_path, MBR = TRUE,
 #' @param global_qvalue_cutoff Global Q-value cutoff
 #' @param qvalue_cutoff Q-value cutoff
 #' @param pg_qvalue_cutoff Protein group Q-value cutoff
-#' @importFrom MSstatsConvert MSstatsImport MSstatsClean
+#' @param annotation Annotation file or data frame
+#' @importFrom MSstatsConvert MSstatsImport MSstatsClean MSstatsMakeAnnotation
 #' @return NULL
 #' @keywords internal
 cleanDIANNChunk = function(input, output_path, MBR, quantificationColumn, pos,
                            global_qvalue_cutoff = 0.01,
                            qvalue_cutoff = 0.01,
-                           pg_qvalue_cutoff = 0.01) {
+                           pg_qvalue_cutoff = 0.01,
+                           annotation = NULL) {
     input = MSstatsImport(list(input = input),
                           "MSstats", "DIANN")
     input = MSstatsClean(
@@ -58,6 +62,7 @@ cleanDIANNChunk = function(input, output_path, MBR, quantificationColumn, pos,
         qvalue_cutoff = qvalue_cutoff,
         pg_qvalue_cutoff = pg_qvalue_cutoff
     )
+    input = MSstatsMakeAnnotation(input, annotation)
     .writeChunkToFile(input, output_path, pos)
     NULL
 }

R/converters.R

@@ -166,6 +166,7 @@ bigSpectronauttoMSstatsFormat <-  function(input_file, output_file_name,
 #' into memory by using dplyr::collect function.
 #'
 bigDIANNtoMSstatsFormat <- function(input_file, 
+                                    annotation = NULL,
                                     output_file_name,
                                     backend,
                                     MBR = TRUE,
@@ -187,7 +188,7 @@ bigDIANNtoMSstatsFormat <- function(input_file,
                  paste0("reduce_output_", output_file_name),
                  MBR,
                  quantificationColumn,
-                 global_qvalue_cutoff, qvalue_cutoff, pg_qvalue_cutoff)
+                 global_qvalue_cutoff, qvalue_cutoff, pg_qvalue_cutoff, annotation)
 
   # Preprocess the cleaned data (feature selection, etc.)
   msstats_data <- MSstatsPreprocessBig(

tests/testthat/test-clean_DIANN.R

@@ -0,0 +1,39 @@
+library(testthat)
+library(mockery)
+
+context("DIANN cleaning")
+
+test_that("cleanDIANNChunk passes annotation to MSstatsMakeAnnotation", {
+  # Prepare data
+  input_chunk <- data.frame(Run = "Run1", Intensity = 100)
+  annotation <- data.frame(Run = "Run1", Condition = "A", BioReplicate = 1)
+
+  # Mocks
+  m_import <- mock(input_chunk)
+  m_clean <- mock(input_chunk)
+  m_annotate <- mock(merge(input_chunk, annotation, by = "Run"))
+  m_write <- mock(NULL)
+
+  stub(cleanDIANNChunk, "MSstatsImport", m_import)
+  stub(cleanDIANNChunk, "MSstatsClean", m_clean)
+  stub(cleanDIANNChunk, "MSstatsMakeAnnotation", m_annotate)
+  stub(cleanDIANNChunk, ".writeChunkToFile", m_write)
+
+  # Execute
+  cleanDIANNChunk(input_chunk, "output.csv", MBR = TRUE, 
+                  quantificationColumn = "Intensity", pos = 1, 
+                  annotation = annotation)
+
+  # Verify
+  expect_called(m_annotate, 1)
+
+  # Check arguments passed to MSstatsMakeAnnotation
+  args <- mock_args(m_annotate)[[1]]
+  expect_equal(args[[1]], input_chunk) # Input from Clean
+  expect_equal(args[[2]], annotation)  # Annotation passed through
+
+  # Check that the result of annotation is passed to write
+  expect_called(m_write, 1)
+  write_args <- mock_args(m_write)[[1]]
+  expect_equal(write_args[[1]], merge(input_chunk, annotation, by = "Run"))
+})