fixed tests

DESCRIPTION

@@ -9,7 +9,7 @@ Description: This is a customized fork of the original work from Tianwei Yu.
         It takes the adaptive processing of LC/MS metabolomics data further
         with focus on high resolution MS for both LC and GC applications.
 Depends: R (>= 3.50), MASS, mzR, splines, doParallel, foreach,
-        snow, dplyr, tidyr, stringr, tibble, tools, arrow
+        snow, dplyr, tidyr, stringr, tibble, tools, arrow, plyr
 biocViews: Technology, MassSpectrometry
 License: GPL-2
 LazyLoad: yes

R/feature.align.R

@@ -30,9 +30,9 @@ create_metadata <- function(sample_grouped, sample_names) {
 #' @export
 create_intensity_row <- function(sample_grouped) {
   sample_grouped %>%
-    group_by(sample_id) %>%
-    summarise(intensity = sum(area)) %>%
-    pivot_wider(names_from = "sample_id", values_from = "intensity")
+    dplyr::group_by(sample_id) %>%
+    dplyr::summarise(intensity = sum(area)) %>%
+    tidyr::pivot_wider(names_from = "sample_id", values_from = "intensity")
 }
 
 #' Compute median RT for each sample
@@ -41,9 +41,9 @@ create_intensity_row <- function(sample_grouped) {
 #' @export
 create_rt_row <- function(sample_grouped) {
   sample_grouped %>%
-    group_by(sample_id) %>%
-    summarise(rt = median(rt)) %>%
-    pivot_wider(names_from = "sample_id", values_from = "rt")
+    dplyr::group_by(sample_id) %>%
+    dplyr::summarise(rt = median(rt)) %>%
+    tidyr::pivot_wider(names_from = "sample_id", values_from = "rt")
 }
 
 #' Create a list containing 3 tibbles: metadata, intensities and RTs.
@@ -161,10 +161,21 @@ comb <- function(x, ...) {
 #' @return Cleaned tibble.
 #' @export
 clean_data_matrix <- function(x, sample_names) {
-  x %>%
+  x <- x %>%
     replace(is.na(.), 0) %>%
-    dplyr::relocate(sample_names) %>%
-    as_tibble()
+    dplyr::relocate(sample_names) |>
+    add_feature_ids()
+  return(x)
+}
+
+#' Add `id` column to a dataframe
+#' @param x A dataframe
+#' @return The same dataframe but with an additional `id` column 
+#' in first place which contains the rownames.
+#' @export
+add_feature_ids <- function(x) {
+  x$id <- as.numeric(rownames(x))
+  return(tibble::as_tibble(x |> dplyr::relocate(id)))
 }
 
 #' Align peaks from spectra into a feature table.
@@ -217,7 +228,7 @@ create_aligned_feature_table <- function(features_table,
 
   aligned_features$intensity <- clean_data_matrix(aligned_features$intensity, sample_names)
   aligned_features$rt <- clean_data_matrix(aligned_features$rt, sample_names)
-  aligned_features$metadata <- as_tibble(aligned_features$metadata)
+  aligned_features$metadata <- add_feature_ids(aligned_features$metadata)
 
   return(aligned_features)
 }

R/utils.R

@@ -8,6 +8,7 @@ register_functions_to_cluster <- function(cluster) {
         'prof.to.features',
         'load.lcms',
         'adaptive.bin',
+        'add_feature_ids',
         'find.turn.point',
         'msExtrema',
         'find_local_maxima',
@@ -113,9 +114,9 @@ load_aligned_features <- function(metadata_file, intensities_file, rt_file, tol_
     tolerances <- arrow::read_parquet(tol_file)
 
     result <- list()
-    result$metadata <- as_tibble(metadata) |> select(-id)
-    result$intensity <- as_tibble(intensities) |> select(-id)
-    result$rt <- as_tibble(rt) |> select(-id)
+    result$metadata <- as_tibble(metadata)
+    result$intensity <- as_tibble(intensities)
+    result$rt <- as_tibble(rt)
     result$mz_tol_relative <- tolerances$mz_tolerance
     result$rt_tol_relative <- tolerances$rt_tolerance
     return(result)

conda/environment.yaml

@@ -18,4 +18,5 @@ dependencies:
   - r-tidyr
   - r-stringr
   - r-tibble
+  - r-plyr
 

tests/testthat/test-feature-align.R

@@ -1,3 +1,9 @@
+update_expected <- function(actual) {
+  arrow::write_parquet(actual$metadata, file.path("..", "testdata", "aligned", "metadata_table.parquet"))
+  arrow::write_parquet(actual$intensity, file.path("..", "testdata", "aligned", "intensity_table.parquet"))
+  arrow::write_parquet(actual$rt, file.path("..", "testdata", "aligned", "rt_table.parquet"))
+}
+
 patrick::with_parameters_test_that(
   "feature.align test",
   {

tests/testthat/test-hybrid.R

@@ -25,6 +25,7 @@ patrick::with_parameters_test_that("basic hybrid test", {
   actual <- as_tibble(result$recovered_feature_sample_table)
   keys <- c("mz", "rt", "sample", "sample_rt", "sample_intensity")
 
+  # arrow::write_parquet(actual, file.path(testdata, "hybrid", paste0(.test_name, "_recovered_feature_sample_table.parquet")))
   expected <- arrow::read_parquet(
     file.path(testdata, "hybrid", paste0(.test_name, "_recovered_feature_sample_table.parquet"))
   )