From 146af3f012b607280b301592452d50e9fecd995e Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jo=C3=A3o=20Cavalcante?= <jvfecav@gmail.com>
Date: Sat, 8 Jun 2024 08:20:33 -0300
Subject: [PATCH] refactor: Change how quast and recombination inputs work
 (#197)

* refactor: Change way of handling IDs in organize_recomb_data

* fix: Remove relative paths from quast input
---
 bin/organize_recomb_data.py      |  6 +++---
 modules/nf-core/quast/main.nf    |  3 ++-
 modules/nf-core/quast/quast.diff | 16 +++++-----------
 subworkflows/local/assemblyqc.nf |  7 +------
 4 files changed, 11 insertions(+), 21 deletions(-)

diff --git a/bin/organize_recomb_data.py b/bin/organize_recomb_data.py
index aeb5846d..1841afe5 100755
--- a/bin/organize_recomb_data.py
+++ b/bin/organize_recomb_data.py
@@ -35,7 +35,7 @@ def remove_after_second_underscore(val):
     parts = val.split('_')
     if len(parts) > 2:
         return '_'.join(parts[:2])
-    return val
+    return re.sub("\.(.*)|_T1|$", "", val)
 
 def create_recomb_input(quast_report, poppunk_clusters, assembly_samplesheet, file_out):
     # Parsing datasets
@@ -46,13 +46,13 @@ def create_recomb_input(quast_report, poppunk_clusters, assembly_samplesheet, fi
             )
 
     poppunk = read_csv(poppunk_clusters)
-    poppunk["Taxon"] = poppunk["Taxon"].str.replace("\.(.*)|_T1|$", "", regex=True)
+    poppunk["Taxon"] = poppunk["Taxon"].apply(remove_after_second_underscore)
 
     assemblies = read_csv(assembly_samplesheet, names=["id", "assembly_path"])
     assemblies["assembly_path"] = [
         Path(path).name for path in assemblies["assembly_path"].to_list()
     ]
-    assemblies["id"] = assemblies["id"].str.replace("\.(.*)|_T1|$", "", regex=True)
+    assemblies["id"] = assemblies["id"].apply(remove_after_second_underscore)
 
     # Merging datasets
     merged = poppunk.merge(quast, right_on="Assembly", left_on="Taxon").loc[
diff --git a/modules/nf-core/quast/main.nf b/modules/nf-core/quast/main.nf
index fd5958fe..4591a638 100644
--- a/modules/nf-core/quast/main.nf
+++ b/modules/nf-core/quast/main.nf
@@ -13,6 +13,7 @@ process QUAST {
     val use_fasta
     val use_gff
 
+
     output:
     path "${prefix}"    , emit: results
     path "*.tsv"        , emit: tsv
@@ -34,7 +35,7 @@ process QUAST {
         $features \\
         --threads $task.cpus \\
         $args \\
-        \$(cat ${consensus.join(' ')})
+        ${consensus.join(' ')}
 
     ln -s ${prefix}/report.tsv
 
diff --git a/modules/nf-core/quast/quast.diff b/modules/nf-core/quast/quast.diff
index 7818dea1..b26c9950 100644
--- a/modules/nf-core/quast/quast.diff
+++ b/modules/nf-core/quast/quast.diff
@@ -10,8 +10,11 @@ Changes in module 'nf-core/quast'
      container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
          'https://depot.galaxyproject.org/singularity/quast:5.2.0--py39pl5321h2add14b_1' :
          'quay.io/biocontainers/quast:5.2.0--py39pl5321h2add14b_1' }"
-@@ -15,7 +15,8 @@
+@@ -13,9 +13,11 @@
+     val use_fasta
+     val use_gff
  
++
      output:
      path "${prefix}"    , emit: results
 -    path '*.tsv'        , emit: tsv
@@ -20,16 +23,7 @@ Changes in module 'nf-core/quast'
      path "versions.yml" , emit: versions
  
      when:
-@@ -33,7 +34,7 @@
-         $features \\
-         --threads $task.cpus \\
-         $args \\
--        ${consensus.join(' ')}
-+        \$(cat ${consensus.join(' ')})
- 
-     ln -s ${prefix}/report.tsv
- 
-@@ -42,4 +43,16 @@
+@@ -42,4 +44,16 @@
          quast: \$(quast.py --version 2>&1 | sed 's/^.*QUAST v//; s/ .*\$//')
      END_VERSIONS
      """
diff --git a/subworkflows/local/assemblyqc.nf b/subworkflows/local/assemblyqc.nf
index 65c6e8eb..6d48d02e 100644
--- a/subworkflows/local/assemblyqc.nf
+++ b/subworkflows/local/assemblyqc.nf
@@ -27,13 +27,8 @@ workflow CHECK_ASSEMBLIES {
         /*
         * Module: QUAST quality check
         */
-        assemblies
-            .map { meta, fasta -> fasta.toString() }
-            .collectFile(name:'assemblies.txt', newLine: true)
-            .set { quast_input }
-
         QUAST (
-            quast_input,
+            assemblies.collect { meta, fasta -> fasta },
             reference_genome,
             [],
             use_reference_genome,