From 1728df69ccbaf4ccb1027c6fd4e9191f48c22194 Mon Sep 17 00:00:00 2001
From: Evangelos Karatzas <32259775+vagkaratzas@users.noreply.github.com>
Date: Thu, 28 Sep 2023 11:39:35 +0100
Subject: [PATCH] Blast (#3889)

* blast suite container version update + stubs

* tests updated and reformatted
---
 modules/nf-core/blast/blastn/main.nf          | 25 +++++++++++----
 modules/nf-core/blast/blastn/meta.yml         |  3 +-
 modules/nf-core/blast/blastp/main.nf          |  2 +-
 modules/nf-core/blast/makeblastdb/main.nf     | 32 ++++++++++++++++---
 modules/nf-core/blast/makeblastdb/meta.yml    |  1 +
 modules/nf-core/blast/tblastn/main.nf         | 25 +++++++++++----
 modules/nf-core/blast/tblastn/meta.yml        |  3 +-
 tests/modules/nf-core/blast/blastn/main.nf    |  1 +
 tests/modules/nf-core/blast/blastn/test.yml   | 22 +++++++------
 tests/modules/nf-core/blast/blastp/test.yml   | 24 +++++++-------
 .../modules/nf-core/blast/makeblastdb/main.nf |  1 +
 .../nf-core/blast/makeblastdb/test.yml        | 18 +++++------
 tests/modules/nf-core/blast/tblastn/main.nf   |  2 +-
 tests/modules/nf-core/blast/tblastn/test.yml  | 22 ++++++-------
 14 files changed, 117 insertions(+), 64 deletions(-)

diff --git a/modules/nf-core/blast/blastn/main.nf b/modules/nf-core/blast/blastn/main.nf
index 9a1f3a55b71..85a54cd1ccb 100644
--- a/modules/nf-core/blast/blastn/main.nf
+++ b/modules/nf-core/blast/blastn/main.nf
@@ -2,18 +2,18 @@ process BLAST_BLASTN {
     tag "$meta.id"
     label 'process_medium'
 
-    conda "bioconda::blast=2.13.0"
+    conda "bioconda::blast=2.14.1"
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'https://depot.galaxyproject.org/singularity/blast:2.13.0--hf3cf87c_0' :
-        'biocontainers/blast:2.13.0--hf3cf87c_0' }"
+        'https://depot.galaxyproject.org/singularity/blast:2.14.1--pl5321h6f7f691_0':
+        'biocontainers/blast:2.14.1--pl5321h6f7f691_0' }"
 
     input:
     tuple val(meta), path(fasta)
     path  db
 
     output:
-    tuple val(meta), path('*.blastn.txt'), emit: txt
-    path "versions.yml"                  , emit: versions
+    tuple val(meta), path('*.txt'), emit: txt
+    path "versions.yml"           , emit: versions
 
     when:
     task.ext.when == null || task.ext.when
@@ -28,7 +28,20 @@ process BLAST_BLASTN {
         -db \$DB \\
         -query $fasta \\
         $args \\
-        -out ${prefix}.blastn.txt
+        -out ${prefix}.txt
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        blast: \$(blastn -version 2>&1 | sed 's/^.*blastn: //; s/ .*\$//')
+    END_VERSIONS
+    """
+
+    stub:
+    def args = task.ext.args ?: ''
+    def prefix = task.ext.prefix ?: "${meta.id}"
+    """
+    touch ${prefix}.txt
+
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
         blast: \$(blastn -version 2>&1 | sed 's/^.*blastn: //; s/ .*\$//')
diff --git a/modules/nf-core/blast/blastn/meta.yml b/modules/nf-core/blast/blastn/meta.yml
index 2742278d511..e43dcce98cd 100644
--- a/modules/nf-core/blast/blastn/meta.yml
+++ b/modules/nf-core/blast/blastn/meta.yml
@@ -31,7 +31,7 @@ output:
   - txt:
       type: file
       description: File containing blastn hits
-      pattern: "*.{blastn.txt}"
+      pattern: "*.txt"
   - versions:
       type: file
       description: File containing software versions
@@ -39,3 +39,4 @@ output:
 authors:
   - "@joseespinosa"
   - "@drpatelh"
+  - "@vagkaratzas"
diff --git a/modules/nf-core/blast/blastp/main.nf b/modules/nf-core/blast/blastp/main.nf
index 1c94a3e6bb7..8f46dcb8cfc 100644
--- a/modules/nf-core/blast/blastp/main.nf
+++ b/modules/nf-core/blast/blastp/main.nf
@@ -13,7 +13,7 @@ process BLAST_BLASTP {
 
     output:
     tuple val(meta), path("*.csv"), emit: csv
-    path "versions.yml"                  , emit: versions
+    path "versions.yml"           , emit: versions
 
     when:
     task.ext.when == null || task.ext.when
diff --git a/modules/nf-core/blast/makeblastdb/main.nf b/modules/nf-core/blast/makeblastdb/main.nf
index f57636116fb..fe705b3f9f4 100644
--- a/modules/nf-core/blast/makeblastdb/main.nf
+++ b/modules/nf-core/blast/makeblastdb/main.nf
@@ -2,17 +2,17 @@ process BLAST_MAKEBLASTDB {
     tag "$fasta"
     label 'process_medium'
 
-    conda "bioconda::blast=2.13.0"
+    conda "bioconda::blast=2.14.1"
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'https://depot.galaxyproject.org/singularity/blast:2.13.0--hf3cf87c_0' :
-        'biocontainers/blast:2.13.0--hf3cf87c_0' }"
+        'https://depot.galaxyproject.org/singularity/blast:2.14.1--pl5321h6f7f691_0':
+        'biocontainers/blast:2.14.1--pl5321h6f7f691_0' }"
 
     input:
     path fasta
 
     output:
-    path 'blast_db'     , emit: db
-    path "versions.yml" , emit: versions
+    path 'blast_db'    , emit: db
+    path "versions.yml", emit: versions
 
     when:
     task.ext.when == null || task.ext.when
@@ -25,6 +25,28 @@ process BLAST_MAKEBLASTDB {
         $args
     mkdir blast_db
     mv ${fasta}* blast_db
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        blast: \$(blastn -version 2>&1 | sed 's/^.*blastn: //; s/ .*\$//')
+    END_VERSIONS
+    """
+
+    stub:
+    def args = task.ext.args ?: ''
+    """
+    touch ${fasta}.fasta
+    touch ${fasta}.fasta.ndb
+    touch ${fasta}.fasta.nhr
+    touch ${fasta}.fasta.nin
+    touch ${fasta}.fasta.njs
+    touch ${fasta}.fasta.not
+    touch ${fasta}.fasta.nsq
+    touch ${fasta}.fasta.ntf
+    touch ${fasta}.fasta.nto
+    mkdir blast_db
+    mv ${fasta}* blast_db
+
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
         blast: \$(blastn -version 2>&1 | sed 's/^.*blastn: //; s/ .*\$//')
diff --git a/modules/nf-core/blast/makeblastdb/meta.yml b/modules/nf-core/blast/makeblastdb/meta.yml
index 83c4b292713..272798d20f4 100644
--- a/modules/nf-core/blast/makeblastdb/meta.yml
+++ b/modules/nf-core/blast/makeblastdb/meta.yml
@@ -29,3 +29,4 @@ output:
 authors:
   - "@joseespinosa"
   - "@drpatelh"
+  - "@vagkaratzas"
diff --git a/modules/nf-core/blast/tblastn/main.nf b/modules/nf-core/blast/tblastn/main.nf
index c8946532c27..58f6ea79851 100644
--- a/modules/nf-core/blast/tblastn/main.nf
+++ b/modules/nf-core/blast/tblastn/main.nf
@@ -2,18 +2,18 @@ process BLAST_TBLASTN {
     tag "$meta.id"
     label 'process_medium'
 
-    conda "bioconda::blast=2.13.0"
+    conda "bioconda::blast=2.14.1"
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'https://depot.galaxyproject.org/singularity/blast:2.13.0--hf3cf87c_0' :
-        'biocontainers/blast:2.13.0--hf3cf87c_0' }"
+        'https://depot.galaxyproject.org/singularity/blast:2.14.1--pl5321h6f7f691_0':
+        'biocontainers/blast:2.14.1--pl5321h6f7f691_0' }"
 
     input:
     tuple val(meta), path(fasta)
     path  db
 
     output:
-    tuple val(meta), path('*.tblastn.txt') , emit: txt
-    path "versions.yml"                    , emit: versions
+    tuple val(meta), path('*.txt'), emit: txt
+    path "versions.yml"           , emit: versions
 
     when:
     task.ext.when == null || task.ext.when
@@ -28,7 +28,20 @@ process BLAST_TBLASTN {
         -db \$DB \\
         -query $fasta \\
         $args \\
-        -out ${prefix}.tblastn.txt
+        -out ${prefix}.txt
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        blast: \$(tblastn -version 2>&1 | sed 's/^.*tblastn: //; s/ .*\$//')
+    END_VERSIONS
+    """
+
+    stub:
+    def args = task.ext.args ?: ''
+    def prefix = task.ext.prefix ?: "${meta.id}"
+    """
+    touch ${prefix}.txt
+
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
         blast: \$(tblastn -version 2>&1 | sed 's/^.*tblastn: //; s/ .*\$//')
diff --git a/modules/nf-core/blast/tblastn/meta.yml b/modules/nf-core/blast/tblastn/meta.yml
index 2359cae6b7b..ae9d003b3ab 100644
--- a/modules/nf-core/blast/tblastn/meta.yml
+++ b/modules/nf-core/blast/tblastn/meta.yml
@@ -31,7 +31,7 @@ output:
   - txt:
       type: file
       description: File containing blastn hits
-      pattern: "*.{tblastn.txt}"
+      pattern: "*.txt"
   - versions:
       type: file
       description: File containing software versions
@@ -39,3 +39,4 @@ output:
 authors:
   - "@yumisims"
   - "@gq2"
+  - "@vagkaratzas"
diff --git a/tests/modules/nf-core/blast/blastn/main.nf b/tests/modules/nf-core/blast/blastn/main.nf
index cb0f8bdf52c..1d1f572cab9 100644
--- a/tests/modules/nf-core/blast/blastn/main.nf
+++ b/tests/modules/nf-core/blast/blastn/main.nf
@@ -6,6 +6,7 @@ include { BLAST_MAKEBLASTDB } from '../../../../../modules/nf-core/blast/makebla
 include { BLAST_BLASTN      } from '../../../../../modules/nf-core/blast/blastn/main.nf'
 
 workflow test_blast_blastn {
+
     input = [ file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true) ]
 
     BLAST_MAKEBLASTDB ( input )
diff --git a/tests/modules/nf-core/blast/blastn/test.yml b/tests/modules/nf-core/blast/blastn/test.yml
index 2c3492a1b07..388ba22ec32 100644
--- a/tests/modules/nf-core/blast/blastn/test.yml
+++ b/tests/modules/nf-core/blast/blastn/test.yml
@@ -1,22 +1,24 @@
-- name: blast_blastn
+- name: blast blastn
   command: nextflow run ./tests/modules/nf-core/blast/blastn -entry test_blast_blastn -c ./tests/config/nextflow.config -c ./tests/modules/nf-core/blast/blastn/nextflow.config
   tags:
     - blast
     - blast/blastn
   files:
-    - path: ./output/blast/test.blastn.txt
-    - path: ./output/blast/blast_db/genome.fasta.nsq
+    - path: output/blast/test.txt
+      contains:
+        - "Query= MT192765.1 Severe acute respiratory syndrome coronavirus 2 isolate"
+    - path: output/blast/blast_db/genome.fasta.nsq
       md5sum: 982cbc7d9e38743b9b1037588862b9da
-    - path: ./output/blast/blast_db/genome.fasta.nin
-    - path: ./output/blast/blast_db/genome.fasta
+    - path: output/blast/blast_db/genome.fasta.nin
+    - path: output/blast/blast_db/genome.fasta
       md5sum: 6e9fe4042a72f2345f644f239272b7e6
-    - path: ./output/blast/blast_db/genome.fasta.nhr
+    - path: output/blast/blast_db/genome.fasta.nhr
       md5sum: f4b4ddb034fd3dd7b25c89e9d50c004e
-    - path: ./output/blast/blast_db/genome.fasta.ndb
+    - path: output/blast/blast_db/genome.fasta.ndb
       md5sum: 0d553c830656469211de113c5022f06d
-    - path: ./output/blast/blast_db/genome.fasta.not
+    - path: output/blast/blast_db/genome.fasta.not
       md5sum: 1e53e9d08f1d23af0299cfa87478a7bb
-    - path: ./output/blast/blast_db/genome.fasta.nto
+    - path: output/blast/blast_db/genome.fasta.nto
       md5sum: 33cdeccccebe80329f1fdbee7f5874cb
-    - path: ./output/blast/blast_db/genome.fasta.ntf
+    - path: output/blast/blast_db/genome.fasta.ntf
       md5sum: de1250813f0c7affc6d12dac9d0fb6bb
diff --git a/tests/modules/nf-core/blast/blastp/test.yml b/tests/modules/nf-core/blast/blastp/test.yml
index 02a844a7d06..d5d3eadc4c2 100644
--- a/tests/modules/nf-core/blast/blastp/test.yml
+++ b/tests/modules/nf-core/blast/blastp/test.yml
@@ -1,24 +1,24 @@
-- name: "blast blastp"
+- name: blast blastp
   command: nextflow run ./tests/modules/nf-core/blast/blastp -entry test_blast_blastp -c ./tests/config/nextflow.config
   tags:
-    - "blast"
-    - "blast/blastp"
+    - blast
+    - blast/blastp
   files:
     - path: output/blast/test.csv
       md5sum: 37a8b6ef8942b5d465be26126bce8965
-    - path: ./output/blast/blast_db/genome.fasta
+    - path: output/blast/blast_db/genome.fasta
       md5sum: 6e9fe4042a72f2345f644f239272b7e6
-    - path: ./output/blast/blast_db/genome.fasta.pdb
+    - path: output/blast/blast_db/genome.fasta.pdb
       md5sum: 0d553c830656469211de113c5022f06d
-    - path: ./output/blast/blast_db/genome.fasta.phr
+    - path: output/blast/blast_db/genome.fasta.phr
       md5sum: f4b4ddb034fd3dd7b25c89e9d50c004e
-    - path: ./output/blast/blast_db/genome.fasta.pin
-    - path: ./output/blast/blast_db/genome.fasta.pjs
-    - path: ./output/blast/blast_db/genome.fasta.pot
+    - path: output/blast/blast_db/genome.fasta.pin
+    - path: output/blast/blast_db/genome.fasta.pjs
+    - path: output/blast/blast_db/genome.fasta.pot
       md5sum: 1e53e9d08f1d23af0299cfa87478a7bb
-    - path: ./output/blast/blast_db/genome.fasta.psq
+    - path: output/blast/blast_db/genome.fasta.psq
       md5sum: cf2ba5f8fc9088651c82de851b4c45a4
-    - path: ./output/blast/blast_db/genome.fasta.ptf
+    - path: output/blast/blast_db/genome.fasta.ptf
       md5sum: de1250813f0c7affc6d12dac9d0fb6bb
-    - path: ./output/blast/blast_db/genome.fasta.pto
+    - path: output/blast/blast_db/genome.fasta.pto
       md5sum: 33cdeccccebe80329f1fdbee7f5874cb
diff --git a/tests/modules/nf-core/blast/makeblastdb/main.nf b/tests/modules/nf-core/blast/makeblastdb/main.nf
index 6822a9fb4ab..bce5321b058 100644
--- a/tests/modules/nf-core/blast/makeblastdb/main.nf
+++ b/tests/modules/nf-core/blast/makeblastdb/main.nf
@@ -5,6 +5,7 @@ nextflow.enable.dsl = 2
 include { BLAST_MAKEBLASTDB } from '../../../../../modules/nf-core/blast/makeblastdb/main.nf'
 
 workflow test_blast_makeblastdb {
+
     input =  [ file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true) ]
 
     BLAST_MAKEBLASTDB ( input )
diff --git a/tests/modules/nf-core/blast/makeblastdb/test.yml b/tests/modules/nf-core/blast/makeblastdb/test.yml
index a1f527fb430..d65cb06e9b5 100644
--- a/tests/modules/nf-core/blast/makeblastdb/test.yml
+++ b/tests/modules/nf-core/blast/makeblastdb/test.yml
@@ -1,21 +1,21 @@
-- name: blast_makeblastdb
+- name: blast makeblastdb
   command: nextflow run ./tests/modules/nf-core/blast/makeblastdb -entry test_blast_makeblastdb -c ./tests/config/nextflow.config -c ./tests/modules/nf-core/blast/makeblastdb/nextflow.config
   tags:
     - blast
     - blast/makeblastdb
   files:
-    - path: ./output/blast/blast_db/genome.fasta.nsq
+    - path: output/blast/blast_db/genome.fasta.nsq
       md5sum: 982cbc7d9e38743b9b1037588862b9da
-    - path: ./output/blast/blast_db/genome.fasta.nin
-    - path: ./output/blast/blast_db/genome.fasta
+    - path: output/blast/blast_db/genome.fasta.nin
+    - path: output/blast/blast_db/genome.fasta
       md5sum: 6e9fe4042a72f2345f644f239272b7e6
-    - path: ./output/blast/blast_db/genome.fasta.nhr
+    - path: output/blast/blast_db/genome.fasta.nhr
       md5sum: f4b4ddb034fd3dd7b25c89e9d50c004e
-    - path: ./output/blast/blast_db/genome.fasta.ndb
+    - path: output/blast/blast_db/genome.fasta.ndb
       md5sum: 0d553c830656469211de113c5022f06d
-    - path: ./output/blast/blast_db/genome.fasta.not
+    - path: output/blast/blast_db/genome.fasta.not
       md5sum: 1e53e9d08f1d23af0299cfa87478a7bb
-    - path: ./output/blast/blast_db/genome.fasta.nto
+    - path: output/blast/blast_db/genome.fasta.nto
       md5sum: 33cdeccccebe80329f1fdbee7f5874cb
-    - path: ./output/blast/blast_db/genome.fasta.ntf
+    - path: output/blast/blast_db/genome.fasta.ntf
       md5sum: de1250813f0c7affc6d12dac9d0fb6bb
diff --git a/tests/modules/nf-core/blast/tblastn/main.nf b/tests/modules/nf-core/blast/tblastn/main.nf
index ee40d8386f0..dfcb7b0167b 100644
--- a/tests/modules/nf-core/blast/tblastn/main.nf
+++ b/tests/modules/nf-core/blast/tblastn/main.nf
@@ -2,8 +2,8 @@
 
 nextflow.enable.dsl = 2
 
-include { BLAST_TBLASTN } from '../../../../../modules/nf-core/blast/tblastn/main.nf'
 include { BLAST_MAKEBLASTDB } from '../../../../../modules/nf-core/blast/makeblastdb/main.nf'
+include { BLAST_TBLASTN } from '../../../../../modules/nf-core/blast/tblastn/main.nf'
 
 workflow test_blast_tblastn {
 
diff --git a/tests/modules/nf-core/blast/tblastn/test.yml b/tests/modules/nf-core/blast/tblastn/test.yml
index 625f06dd09c..41badfcb498 100644
--- a/tests/modules/nf-core/blast/tblastn/test.yml
+++ b/tests/modules/nf-core/blast/tblastn/test.yml
@@ -4,22 +4,20 @@
     - blast
     - blast/tblastn
   files:
-    - path: ./output/blast/test.tblastn.txt
-      md5sum: c26ccfa5ceae82fc6affdc77f5505b62
-    - path: ./output/blast/versions.yml
-      md5sum: 63d79c4c09cfd691d6be814e7fa3efd3
-    - path: ./output/blast/blast_db/genome.fasta.nsq
+    - path: output/blast/test.txt
+      md5sum: a78484fae1573991fabf727a7911c571
+    - path: output/blast/blast_db/genome.fasta.nsq
       md5sum: 982cbc7d9e38743b9b1037588862b9da
-    - path: ./output/blast/blast_db/genome.fasta.nin
-    - path: ./output/blast/blast_db/genome.fasta
+    - path: output/blast/blast_db/genome.fasta.nin
+    - path: output/blast/blast_db/genome.fasta
       md5sum: 6e9fe4042a72f2345f644f239272b7e6
-    - path: ./output/blast/blast_db/genome.fasta.nhr
+    - path: output/blast/blast_db/genome.fasta.nhr
       md5sum: f4b4ddb034fd3dd7b25c89e9d50c004e
-    - path: ./output/blast/blast_db/genome.fasta.ndb
+    - path: output/blast/blast_db/genome.fasta.ndb
       md5sum: 0d553c830656469211de113c5022f06d
-    - path: ./output/blast/blast_db/genome.fasta.not
+    - path: output/blast/blast_db/genome.fasta.not
       md5sum: 1e53e9d08f1d23af0299cfa87478a7bb
-    - path: ./output/blast/blast_db/genome.fasta.nto
+    - path: output/blast/blast_db/genome.fasta.nto
       md5sum: 33cdeccccebe80329f1fdbee7f5874cb
-    - path: ./output/blast/blast_db/genome.fasta.ntf
+    - path: output/blast/blast_db/genome.fasta.ntf
       md5sum: de1250813f0c7affc6d12dac9d0fb6bb