From be8528998134132c6f4823f627bbb0aec6a51747 Mon Sep 17 00:00:00 2001
From: "James A. Fellows Yates" <jfy133@gmail.com>
Date: Thu, 9 Dec 2021 15:00:32 +0100
Subject: [PATCH] Update metabat2 output channels and compression (#1111)

* feat: each output type has dedicated channel replace bgzip with gzip can only zip one at a time

* Add condition moving of unbinned files

* fix: solution for moving sometimes non-existant files

* fix: update meta.yml to add the new channels

* fix: remove most of the checksums due to variability

* fix: tweaking of output

* Update modules/metabat2/metabat2/main.nf

Co-authored-by: Maxime Borry <maxibor@users.noreply.github.com>

* Fix find commands

* Fix find commands

Co-authored-by: Maxime Borry <maxibor@users.noreply.github.com>
---
 modules/metabat2/metabat2/main.nf        | 15 ++++++++++-----
 modules/metabat2/metabat2/meta.yml       | 13 +++++++++++++
 tests/modules/metabat2/metabat2/test.yml | 15 ++++++++-------
 3 files changed, 31 insertions(+), 12 deletions(-)

diff --git a/modules/metabat2/metabat2/main.nf b/modules/metabat2/metabat2/main.nf
index 2d01fdf6..a8af0ae9 100644
--- a/modules/metabat2/metabat2/main.nf
+++ b/modules/metabat2/metabat2/main.nf
@@ -11,9 +11,12 @@ process METABAT2_METABAT2 {
     tuple val(meta), path(fasta), path(depth)
 
     output:
-    tuple val(meta), path("bins/*.fa.gz")       , optional:true , emit: fasta
-    tuple val(meta), path("*.tsv.gz"), optional:true , emit: membership
-    path "versions.yml"                                         , emit: versions
+    tuple val(meta), path("*.tooShort.fa.gz")       , optional:true , emit: tooshort
+    tuple val(meta), path("*.lowDepth.fa.gz")       , optional:true , emit: lowdepth
+    tuple val(meta), path("*.unbinned.fa.gz")       , optional:true , emit: unbinned
+    tuple val(meta), path("*.tsv.gz")               , optional:true , emit: membership
+    tuple val(meta), path("bins/*.fa.gz")           , optional:true , emit: fasta
+    path "versions.yml"                                             , emit: versions
 
     script:
     def args = task.ext.args ?: ''
@@ -33,8 +36,10 @@ process METABAT2_METABAT2 {
 
     mv metabat2/${prefix} ${prefix}.tsv
     mv metabat2 bins
-    bgzip --threads $task.cpus ${prefix}.tsv
-    bgzip --threads $task.cpus bins/*.fa
+
+    gzip ${prefix}.tsv
+    find ./bins/ -name "*.fa" -type f | xargs -t -n 1 bgzip -@ ${task.cpus}
+    find ./bins/ -name "*[lowDepth,tooShort,unbinned].fa.gz" -type f -exec mv {} . \\;
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
diff --git a/modules/metabat2/metabat2/meta.yml b/modules/metabat2/metabat2/meta.yml
index a7f3a7ff..0ec07b02 100644
--- a/modules/metabat2/metabat2/meta.yml
+++ b/modules/metabat2/metabat2/meta.yml
@@ -46,6 +46,18 @@ output:
       type: file
       description: Bins created from assembled contigs in fasta file
       pattern: "*.fa.gz"
+  - tooshort:
+      type: file
+      description: Contigs that did not pass length filtering
+      pattern: "*.tooShort.fa.gz"
+  - lowdepth:
+      type: file
+      description: Contigs that did not have sufficient depth for binning
+      pattern: "*.lowDepth.fa.gz"
+  - unbinned:
+      type: file
+      description: Contigs that pass length and depth filtering but could not be binned
+      pattern: "*.unbinned.fa.gz"
   - membership:
       type: file
       description: cluster memberships as a matrix format.
@@ -54,3 +66,4 @@ output:
 
 authors:
   - "@maxibor"
+  - "@jfy133"
diff --git a/tests/modules/metabat2/metabat2/test.yml b/tests/modules/metabat2/metabat2/test.yml
index 1a8660a7..9389295e 100644
--- a/tests/modules/metabat2/metabat2/test.yml
+++ b/tests/modules/metabat2/metabat2/test.yml
@@ -1,23 +1,24 @@
 - name: metabat2 metabat2 test_metabat2_no_depth
-  command: nextflow run ./tests/modules/metabat2/metabat2 -entry test_metabat2_no_depth -c ./tests/config/nextflow.config -c ./tests/modules/metabat2/metabat2/nextflow.config
+  command: nextflow run tests/modules/metabat2/metabat2 -entry test_metabat2_no_depth -c tests/config/nextflow.config
   tags:
-    - metabat2/metabat2
     - metabat2
+    - metabat2/metabat2
   files:
     - path: output/metabat2/bins/test.1.fa.gz
       md5sum: 0e9bce5b5a0033fd4411a21dec881170
     - path: output/metabat2/test.tsv.gz
-      md5sum: ea77e8c4426d2337419905b57f1ec335
+    - path: output/metabat2/versions.yml
+      md5sum: 5742a71af36c3a748fd5726d76924ba8
 
 - name: metabat2 metabat2 test_metabat2_depth
-  command: nextflow run ./tests/modules/metabat2/metabat2 -entry test_metabat2_depth -c ./tests/config/nextflow.config -c ./tests/modules/metabat2/metabat2/nextflow.config
+  command: nextflow run tests/modules/metabat2/metabat2 -entry test_metabat2_depth -c tests/config/nextflow.config
   tags:
-    - metabat2/metabat2
     - metabat2
+    - metabat2/metabat2
   files:
     - path: output/metabat2/bins/test.1.fa.gz
       md5sum: 0e9bce5b5a0033fd4411a21dec881170
     - path: output/metabat2/test.tsv.gz
-      md5sum: ea77e8c4426d2337419905b57f1ec335
     - path: output/metabat2/test.txt.gz
-      md5sum: 8f735aa408d6c90e5a0310e06ace7a9a
+    - path: output/metabat2/versions.yml
+      md5sum: 538c56b2df7d90580f05097218b5d5b1