From e432a05dc59832d55307442264ad2b70039f5e3a Mon Sep 17 00:00:00 2001
From: WackerO <43847497+WackerO@users.noreply.github.com>
Date: Thu, 5 Oct 2023 10:25:19 +0200
Subject: [PATCH] Fix CUSTOM/MATRIXFILTER (#3912)

* Added log_assays param to shinyngs tests, updated shinyngs containers, removed na.omit() from proteus

* Updated shinyngs containers, added test for plot_exploratory
--log_assays param

* added --log2_assays flag back into nextflow.config

* removed redundant process definition

* prettier, updated proteus tests

* Added na.rm to matrixfilter

* Made params more uniform, updated meta.yml and main.nf

* Bugfix, added tests

* linting

* Combined rejection files into one file

* Updated test file path

* Propose more generic framework for matrixfilter

* Update output definitition

* Fix bool output

* Update tests

* appease eclint

* fixed matrixfilter bug, added comments

---------

Co-authored-by: Jonathan Manning <pininforthefjords@gmail.com>
---
 .../nf-core/custom/matrixfilter/templates/matrixfilter.R  | 6 ++++--
 tests/modules/nf-core/custom/matrixfilter/test.yml        | 8 ++++----
 2 files changed, 8 insertions(+), 6 deletions(-)

diff --git a/modules/nf-core/custom/matrixfilter/templates/matrixfilter.R b/modules/nf-core/custom/matrixfilter/templates/matrixfilter.R
index dc8b508f1c2..0bfff8cca41 100644
--- a/modules/nf-core/custom/matrixfilter/templates/matrixfilter.R
+++ b/modules/nf-core/custom/matrixfilter/templates/matrixfilter.R
@@ -156,6 +156,8 @@ if ((opt\$sample_file != '') && ( ! is.null(opt\$grouping_variable))){
 
 # Also set up filtering for NAs; use by default minimum_proportion_not_na; only
 # use minimum_samples_not_na if it is provided (default NULL)
+# -->NA test can always use minimum_samples_not_na as this will contain the correct
+# value even if the proportion is to be used
 
 if (is.null(opt\$minimum_samples_not_na)) {
     opt\$minimum_samples_not_na <- ncol(abundance_matrix) * opt\$minimum_proportion_not_na
@@ -164,8 +166,8 @@ if (is.null(opt\$minimum_samples_not_na)) {
 # Define the tests
 
 tests <- list(
-    'abundance' = function(x) sum(x > opt\$minimum_abundance, na.rm = T) >= opt\$minimum_samples,
-    'na' = function(x) !any(is.na(x)) || sum(!is.na(x))/length(x) >= opt\$minimum_samples_not_n
+    'abundance' = function(x) sum(x > opt\$minimum_abundance, na.rm = T) >= opt\$minimum_samples, # check if rows have sufficiently high abundance
+    'na' = function(x) !any(is.na(x)) || sum(!is.na(x)) >= opt\$minimum_samples_not_na  # check if enough values in row are not NA
 )
 
 # Apply the functions row-wise on the abundance_matrix and store the result in a boolean matrix
diff --git a/tests/modules/nf-core/custom/matrixfilter/test.yml b/tests/modules/nf-core/custom/matrixfilter/test.yml
index cf86a8bacf2..f8af0f09039 100644
--- a/tests/modules/nf-core/custom/matrixfilter/test.yml
+++ b/tests/modules/nf-core/custom/matrixfilter/test.yml
@@ -45,9 +45,9 @@
   files:
     - path: output/custom/R_sessionInfo.log
     - path: output/custom/mq_prop_test.filtered.tsv
-      md5sum: a0943baaa48ea76bdb8c80a580dc9953
+      md5sum: f5bcfe041816e20ef4cd89bbe0a7e7a2
     - path: output/custom/mq_prop_test.tests.tsv
-      md5sum: 65f7c7ad7e475d82985bd3fc4fec0451
+      md5sum: 12fc7c32300f774924c5d4ad59b1e447
     - path: output/custom/versions.yml
 
 - name: custom matrixfilter test_custom_matrixfilter_na_samples
@@ -58,7 +58,7 @@
   files:
     - path: output/custom/R_sessionInfo.log
     - path: output/custom/mq_samples_test.filtered.tsv
-      md5sum: f0d81baf35d7eb324365ad5e1c6eb2ae
+      md5sum: 9deea105a4bf8bf7128a7621d4acedb6
     - path: output/custom/mq_samples_test.tests.tsv
-      md5sum: c9e91fb2e26ba8f5d3c73f0d2252fb56
+      md5sum: be76a1f032f4c7650faa910a87a5467f
     - path: output/custom/versions.yml