From 31767ccf9b5d24286e34718caa9541ccdfa79bdb Mon Sep 17 00:00:00 2001
From: prototaxites <jimdownie@fastmail.org>
Date: Mon, 24 Jul 2023 19:55:25 +0000
Subject: [PATCH 01/19] Add galah

---
 modules/nf-core/galah/main.nf               | 57 ++++++++++++++++++
 modules/nf-core/galah/meta.yml              | 64 +++++++++++++++++++++
 tests/config/pytest_modules.yml             |  4 ++
 tests/modules/nf-core/galah/main.nf         | 21 +++++++
 tests/modules/nf-core/galah/nextflow.config |  5 ++
 tests/modules/nf-core/galah/test.yml        |  8 +++
 6 files changed, 159 insertions(+)
 create mode 100644 modules/nf-core/galah/main.nf
 create mode 100644 modules/nf-core/galah/meta.yml
 create mode 100644 tests/modules/nf-core/galah/main.nf
 create mode 100644 tests/modules/nf-core/galah/nextflow.config
 create mode 100644 tests/modules/nf-core/galah/test.yml

diff --git a/modules/nf-core/galah/main.nf b/modules/nf-core/galah/main.nf
new file mode 100644
index 00000000000..f498663d5a4
--- /dev/null
+++ b/modules/nf-core/galah/main.nf
@@ -0,0 +1,57 @@
+process GALAH {
+    tag "$meta.id"
+    label 'process_medium'
+
+    conda "bioconda::galah=0.3.1"
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/galah:0.3.1--hec16e2b_1':
+        'biocontainers/galah:0.3.1--hec16e2b_1' }"
+
+    input:
+    tuple val(meta), path(bins)
+    path(checkm_tab_table)
+    path(genome_info)
+
+    output:
+    tuple val(meta), path("*.tsv")                    , emit: tsv
+    tuple val(meta), path("${prefix}-dereplicated/*") , emit: dereplicated_bins, optional: true
+    path "versions.yml"                               , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+    prefix = task.ext.prefix ?: "${meta.id}"
+    def qc_input = checkm_tab_table ? "--checkm-tab-table ${checkm_tab_table}" : (genome_info ? "--genome-info ${genome_info}" : "")
+    if (checkm_tab_table && genome_info) { error "genome_info table and checkm_tab_table both provided: please provide one or the other." }
+    """
+    mkdir ${prefix}-dereplicated
+
+    galah cluster \\
+        --threads ${task.cpus} \\
+        --genome-fasta-files ${bins} \\
+        ${qc_input} \\
+        --output-cluster-definition ${prefix}-dereplicated_bins.tsv \\
+        ${args}
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        : \$(galah --version | sed 's/galah //')
+    END_VERSIONS
+    """
+
+    stub:
+    def args = task.ext.args ?: ''
+    prefix = task.ext.prefix ?: "${meta.id}"
+    """
+    mkdir ${prefix}-dereplicated/
+    touch ${prefix}-dereplicated/test.fa
+    touch ${prefix}-dereplicated_bins.tsv
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        : \$(galah --version | sed 's/galah //')
+    END_VERSIONS
+    """
+}
diff --git a/modules/nf-core/galah/meta.yml b/modules/nf-core/galah/meta.yml
new file mode 100644
index 00000000000..15c198424f8
--- /dev/null
+++ b/modules/nf-core/galah/meta.yml
@@ -0,0 +1,64 @@
+---
+# yaml-language-server: $schema=https://raw.githubusercontent.com/nf-core/modules/master/modules/yaml-schema.json
+name: "galah"
+description: Cluster genome FASTA files by average nucleotide identity
+keywords:
+  - genomics
+  - cluster
+  - genome
+  - metagenomics
+tools:
+  - "galah":
+      description: "Galah aims to be a more scalable metagenome assembled genome (MAG) dereplication method."
+      homepage: "https://github.com/wwood/galah"
+      documentation: "https://github.com/wwood/galah"
+      tool_dev_url: "https://github.com/wwood/galah"
+      doi: ""
+      licence: "['GPL v3']"
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. `[ id:'test', single_end:false ]`
+  - bins:
+      type: file
+      description: A list of fasta-formatted genomes for dereplication
+      pattern: "*.{fa,fna,fa.gz, etc}"
+  - checkm_tab_table:
+      type: file
+      description: |
+        (optional) (CheckM)[https://nf-co.re/modules/checkm_lineagewf] summary TSV containing information on the completeness and contamination of the input genomes.
+        Conflicts with genome_info.
+      pattern: "*.tsv"
+  - genome_info:
+      type: file
+      description: |
+        (optional) CSV file with the header genome,completeness,contamination.
+        `genome` is the name of the genome file, minus the file extension, and `completeness` and
+        `contamination` are the completeness and contamination of each genome in percentages (0-100).
+        Alternative to checkm_tab_table if genome QC scores come from another source (e.g. BUSCO).
+        Conflicts with checkm_tab_table.
+      pattern: "*.csv"
+
+output:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. `[ id:'test', single_end:false ]`
+  - tsv:
+      type: file
+      description: TSV file in the format `representative_genome` \t `member_genome`
+      pattern: "*.tsv"
+  - dereplicated_bins:
+      type: file
+      description: (optional, depending on input args) The representative genomes following dereplication by galah.
+      pattern: "*"
+  - versions:
+      type: file
+      description: File containing software versions
+      pattern: "versions.yml"
+authors:
+  - "@prototaxites"
diff --git a/tests/config/pytest_modules.yml b/tests/config/pytest_modules.yml
index 39a14960902..5e8a442b579 100644
--- a/tests/config/pytest_modules.yml
+++ b/tests/config/pytest_modules.yml
@@ -1243,6 +1243,10 @@ freyja/variants:
   - modules/nf-core/freyja/variants/**
   - tests/modules/nf-core/freyja/variants/**
 
+galah:
+  - modules/nf-core/galah/**
+  - tests/modules/nf-core/galah/**
+
 gamma/gamma:
   - modules/nf-core/gamma/gamma/**
   - tests/modules/nf-core/gamma/gamma/**
diff --git a/tests/modules/nf-core/galah/main.nf b/tests/modules/nf-core/galah/main.nf
new file mode 100644
index 00000000000..e01e2912295
--- /dev/null
+++ b/tests/modules/nf-core/galah/main.nf
@@ -0,0 +1,21 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { GALAH                       } from '../../../../modules/nf-core/galah/main.nf'
+include { BIOAWK as BIOAWK_CHECKM     } from '../../../../modules/nf-core/bioawk/main.nf'
+include { BIOAWK as BIOAWK_GENOMEINFO } from '../../../../modules/nf-core/bioawk/main.nf'
+include { GUNZIP                      } from '../../../../modules/nf-core/gunzip/main.nf'
+
+
+workflow test_galah {
+
+    input = [
+        [ id:'test' ], // meta map
+        [file("https://github.com/nf-core/test-datasets/raw/magmap/testdata/GCA_002688505.1_ASM268850v1_genomic.fna.gz", checkIfExists: true),
+        file("https://github.com/nf-core/test-datasets/raw/magmap/testdata/GCF_004296495.1_ASM429649v1_genomic.fna.gz", checkIfExists: true)]
+    ]
+
+    GALAH ( input, [], [] )
+
+}
diff --git a/tests/modules/nf-core/galah/nextflow.config b/tests/modules/nf-core/galah/nextflow.config
new file mode 100644
index 00000000000..8730f1c4b93
--- /dev/null
+++ b/tests/modules/nf-core/galah/nextflow.config
@@ -0,0 +1,5 @@
+process {
+
+    publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
+
+}
diff --git a/tests/modules/nf-core/galah/test.yml b/tests/modules/nf-core/galah/test.yml
new file mode 100644
index 00000000000..674af2d80ff
--- /dev/null
+++ b/tests/modules/nf-core/galah/test.yml
@@ -0,0 +1,8 @@
+- name: galah test_galah
+  command: nextflow run ./tests/modules/nf-core/galah -entry test_galah -c ./tests/config/nextflow.config -c ./tests/modules/nf-core/galah/nextflow.config
+  tags:
+    - galah
+  files:
+    - path: output/galah/test-dereplicated_bins.tsv
+      md5sum: d2f8a621bfa5794467f4fdd759e2bce7
+    - path: output/galah/versions.yml

From b066cc2c3350c7c263cf63e03510b8a3caef6b22 Mon Sep 17 00:00:00 2001
From: prototaxites <jimdownie@fastmail.org>
Date: Mon, 24 Jul 2023 20:07:19 +0000
Subject: [PATCH 02/19] NO DOI

---
 modules/nf-core/galah/meta.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/nf-core/galah/meta.yml b/modules/nf-core/galah/meta.yml
index 15c198424f8..73ac4b4c21b 100644
--- a/modules/nf-core/galah/meta.yml
+++ b/modules/nf-core/galah/meta.yml
@@ -13,7 +13,7 @@ tools:
       homepage: "https://github.com/wwood/galah"
       documentation: "https://github.com/wwood/galah"
       tool_dev_url: "https://github.com/wwood/galah"
-      doi: ""
+      doi: "10.1111/XXXX"
       licence: "['GPL v3']"
 
 input:

From 65079b563d4d3ddec6ddcd513e0327890106d93d Mon Sep 17 00:00:00 2001
From: prototaxites <jimdownie@fastmail.org>
Date: Mon, 24 Jul 2023 20:10:53 +0000
Subject: [PATCH 03/19] fix versions

---
 modules/nf-core/galah/main.nf | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/nf-core/galah/main.nf b/modules/nf-core/galah/main.nf
index f498663d5a4..1c18dd2cbf8 100644
--- a/modules/nf-core/galah/main.nf
+++ b/modules/nf-core/galah/main.nf
@@ -37,7 +37,7 @@ process GALAH {
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
-        : \$(galah --version | sed 's/galah //')
+        galah: \$(galah --version | sed 's/galah //')
     END_VERSIONS
     """
 
@@ -51,7 +51,7 @@ process GALAH {
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
-        : \$(galah --version | sed 's/galah //')
+        galah: \$(galah --version | sed 's/galah //')
     END_VERSIONS
     """
 }

From c7f9abdaf62abeff139402ca76f604f39828bf72 Mon Sep 17 00:00:00 2001
From: prototaxites <jimdownie@fastmail.org>
Date: Tue, 25 Jul 2023 07:33:14 +0000
Subject: [PATCH 04/19] fix dashing version in conda

---
 modules/nf-core/galah/main.nf | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/nf-core/galah/main.nf b/modules/nf-core/galah/main.nf
index 1c18dd2cbf8..ae5f849ad03 100644
--- a/modules/nf-core/galah/main.nf
+++ b/modules/nf-core/galah/main.nf
@@ -2,7 +2,7 @@ process GALAH {
     tag "$meta.id"
     label 'process_medium'
 
-    conda "bioconda::galah=0.3.1"
+    conda "bioconda::galah=0.3.1 bioconda::dashing=0.4.0-3"
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
         'https://depot.galaxyproject.org/singularity/galah:0.3.1--hec16e2b_1':
         'biocontainers/galah:0.3.1--hec16e2b_1' }"

From 5e4c0af1cea9d59fcf94b0447dad709874d33b8e Mon Sep 17 00:00:00 2001
From: prototaxites <jimdownie@fastmail.org>
Date: Tue, 25 Jul 2023 07:37:26 +0000
Subject: [PATCH 05/19] fix inputs/outputs

---
 modules/nf-core/galah/main.nf  | 4 ++--
 modules/nf-core/galah/meta.yml | 2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/modules/nf-core/galah/main.nf b/modules/nf-core/galah/main.nf
index ae5f849ad03..f1c62924a63 100644
--- a/modules/nf-core/galah/main.nf
+++ b/modules/nf-core/galah/main.nf
@@ -14,7 +14,7 @@ process GALAH {
 
     output:
     tuple val(meta), path("*.tsv")                    , emit: tsv
-    tuple val(meta), path("${prefix}-dereplicated/*") , emit: dereplicated_bins, optional: true
+    tuple val(meta), path("${prefix}-dereplicated/*") , emit: dereplicated_bins
     path "versions.yml"                               , emit: versions
 
     when:
@@ -33,7 +33,7 @@ process GALAH {
         --genome-fasta-files ${bins} \\
         ${qc_input} \\
         --output-cluster-definition ${prefix}-dereplicated_bins.tsv \\
-        ${args}
+        --output-representative-fasta-directory ${prefix}-dereplicated
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
diff --git a/modules/nf-core/galah/meta.yml b/modules/nf-core/galah/meta.yml
index 73ac4b4c21b..c999cd0395d 100644
--- a/modules/nf-core/galah/meta.yml
+++ b/modules/nf-core/galah/meta.yml
@@ -54,7 +54,7 @@ output:
       pattern: "*.tsv"
   - dereplicated_bins:
       type: file
-      description: (optional, depending on input args) The representative genomes following dereplication by galah.
+      description: The representative genomes following dereplication by galah.
       pattern: "*"
   - versions:
       type: file

From 5b672307a637642fe29db240fa43620e0b7151fc Mon Sep 17 00:00:00 2001
From: prototaxites <jimdownie@fastmail.org>
Date: Tue, 25 Jul 2023 07:46:59 +0000
Subject: [PATCH 06/19] fix dashing version

---
 modules/nf-core/galah/main.nf | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/nf-core/galah/main.nf b/modules/nf-core/galah/main.nf
index f1c62924a63..006914df4e2 100644
--- a/modules/nf-core/galah/main.nf
+++ b/modules/nf-core/galah/main.nf
@@ -2,7 +2,7 @@ process GALAH {
     tag "$meta.id"
     label 'process_medium'
 
-    conda "bioconda::galah=0.3.1 bioconda::dashing=0.4.0-3"
+    conda "bioconda::galah=0.3.1 bioconda::dashing=0.4.0"
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
         'https://depot.galaxyproject.org/singularity/galah:0.3.1--hec16e2b_1':
         'biocontainers/galah:0.3.1--hec16e2b_1' }"

From f3feb1cefd135bd06ddabd3dce5df52ac436f566 Mon Sep 17 00:00:00 2001
From: prototaxites <jimdownie@fastmail.org>
Date: Tue, 25 Jul 2023 18:09:36 +0000
Subject: [PATCH 07/19] Update model inputs, add tests

---
 modules/nf-core/galah/main.nf               | 10 ++++----
 modules/nf-core/galah/meta.yml              | 27 ++++++++++----------
 tests/modules/nf-core/galah/main.nf         | 28 ++++++++++++++++++---
 tests/modules/nf-core/galah/nextflow.config |  7 ++++++
 tests/modules/nf-core/galah/test.yml        | 23 +++++++++++++++++
 5 files changed, 72 insertions(+), 23 deletions(-)

diff --git a/modules/nf-core/galah/main.nf b/modules/nf-core/galah/main.nf
index 006914df4e2..6effe726fb4 100644
--- a/modules/nf-core/galah/main.nf
+++ b/modules/nf-core/galah/main.nf
@@ -2,15 +2,15 @@ process GALAH {
     tag "$meta.id"
     label 'process_medium'
 
-    conda "bioconda::galah=0.3.1 bioconda::dashing=0.4.0"
+    conda "bioconda::galah=0.3.1 bioconda::dashing=0.4.0 bioconda::fastani=1.31"
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
         'https://depot.galaxyproject.org/singularity/galah:0.3.1--hec16e2b_1':
         'biocontainers/galah:0.3.1--hec16e2b_1' }"
 
     input:
     tuple val(meta), path(bins)
-    path(checkm_tab_table)
-    path(genome_info)
+    path(qc_table)
+    val(qc_format)
 
     output:
     tuple val(meta), path("*.tsv")                    , emit: tsv
@@ -23,8 +23,8 @@ process GALAH {
     script:
     def args = task.ext.args ?: ''
     prefix = task.ext.prefix ?: "${meta.id}"
-    def qc_input = checkm_tab_table ? "--checkm-tab-table ${checkm_tab_table}" : (genome_info ? "--genome-info ${genome_info}" : "")
-    if (checkm_tab_table && genome_info) { error "genome_info table and checkm_tab_table both provided: please provide one or the other." }
+    def qc_input = qc_table ? ((qc_format == "checkm") ? "--checkm-tab-table ${qc_table}" : "--genome-info ${qc_table}") : ""
+    if( qc_table && !(qc_format in ["checkm", "genome_info"]) ) error "Invalid qc_format supplied!"
     """
     mkdir ${prefix}-dereplicated
 
diff --git a/modules/nf-core/galah/meta.yml b/modules/nf-core/galah/meta.yml
index c999cd0395d..85263a8f591 100644
--- a/modules/nf-core/galah/meta.yml
+++ b/modules/nf-core/galah/meta.yml
@@ -13,7 +13,7 @@ tools:
       homepage: "https://github.com/wwood/galah"
       documentation: "https://github.com/wwood/galah"
       tool_dev_url: "https://github.com/wwood/galah"
-      doi: "10.1111/XXXX"
+      doi: "10.1111/NODOI"
       licence: "['GPL v3']"
 
 input:
@@ -26,21 +26,20 @@ input:
       type: file
       description: A list of fasta-formatted genomes for dereplication
       pattern: "*.{fa,fna,fa.gz, etc}"
-  - checkm_tab_table:
+  - qc_table:
       type: file
       description: |
-        (optional) (CheckM)[https://nf-co.re/modules/checkm_lineagewf] summary TSV containing information on the completeness and contamination of the input genomes.
-        Conflicts with genome_info.
-      pattern: "*.tsv"
-  - genome_info:
-      type: file
-      description: |
-        (optional) CSV file with the header genome,completeness,contamination.
-        `genome` is the name of the genome file, minus the file extension, and `completeness` and
-        `contamination` are the completeness and contamination of each genome in percentages (0-100).
-        Alternative to checkm_tab_table if genome QC scores come from another source (e.g. BUSCO).
-        Conflicts with checkm_tab_table.
-      pattern: "*.csv"
+        (optional) Either a (CheckM)[https://nf-co.re/modules/checkm_lineagewf] summary TSV containing
+        information on the completeness and contamination of the input genomes (13 columns),
+        or a 3-column csv with the header `genome,completeness,contamination`.
+        In both cases the first column should contain the names of the input genome files,
+        minus the last file extension
+        (i.e. if the genome is gzipped, the genome name should retain the .fasta extension).
+      pattern: "*.{csv,tsv}"
+  - qc_format:
+      type: string
+      description: Defines the type if input table in `qc_table`, if specified.
+      pattern: "checkm|genome_info"
 
 output:
   - meta:
diff --git a/tests/modules/nf-core/galah/main.nf b/tests/modules/nf-core/galah/main.nf
index e01e2912295..d260cc67ed0 100644
--- a/tests/modules/nf-core/galah/main.nf
+++ b/tests/modules/nf-core/galah/main.nf
@@ -1,13 +1,9 @@
-#!/usr/bin/env nextflow
-
-nextflow.enable.dsl = 2
 
 include { GALAH                       } from '../../../../modules/nf-core/galah/main.nf'
 include { BIOAWK as BIOAWK_CHECKM     } from '../../../../modules/nf-core/bioawk/main.nf'
 include { BIOAWK as BIOAWK_GENOMEINFO } from '../../../../modules/nf-core/bioawk/main.nf'
 include { GUNZIP                      } from '../../../../modules/nf-core/gunzip/main.nf'
 
-
 workflow test_galah {
 
     input = [
@@ -19,3 +15,27 @@ workflow test_galah {
     GALAH ( input, [], [] )
 
 }
+
+workflow test_galah_genomeinfo {
+
+    input = [
+        [ id:'test' ], // meta map
+        [file("https://github.com/nf-core/test-datasets/raw/magmap/testdata/GCA_002688505.1_ASM268850v1_genomic.fna.gz", checkIfExists: true),
+        file("https://github.com/nf-core/test-datasets/raw/magmap/testdata/GCF_004296495.1_ASM429649v1_genomic.fna.gz", checkIfExists: true)]
+    ]
+
+    genomeinfo = [
+        [ id: 'genomeinfo' ],
+        file("https://raw.githubusercontent.com/nf-core/test-datasets/magmap/testdata/checkm.lineage_wf.qa_2.tsv", checkIfExists: true)
+    ]
+
+    BIOAWK_GENOMEINFO(genomeinfo)
+
+    GUNZIP(BIOAWK_GENOMEINFO.out.output)
+    
+    ch_genomeinfo = GUNZIP.out.gunzip
+        .map { meta, tsv -> [tsv] }
+
+    GALAH ( input, ch_genomeinfo, "genome_info")
+
+}
diff --git a/tests/modules/nf-core/galah/nextflow.config b/tests/modules/nf-core/galah/nextflow.config
index 8730f1c4b93..8eacddfbc81 100644
--- a/tests/modules/nf-core/galah/nextflow.config
+++ b/tests/modules/nf-core/galah/nextflow.config
@@ -2,4 +2,11 @@ process {
 
     publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
 
+    // write a horrid awk to munge the checkm_qa tsv to a 3 column csv
+    // as files are gzipped, genome has to include extension "fna" as galah expects the genome name to be
+    // the file name minus the last extension
+    withName: BIOAWK_GENOMEINFO {
+        ext.args = '\'BEGIN {{FS="\t"; OFS=","}} NR==1 {print "genome","completeness","contamination"} NR>1 {print $1".fna",$6, $7}\''
+    }
+
 }
diff --git a/tests/modules/nf-core/galah/test.yml b/tests/modules/nf-core/galah/test.yml
index 674af2d80ff..9eec3cc7eb7 100644
--- a/tests/modules/nf-core/galah/test.yml
+++ b/tests/modules/nf-core/galah/test.yml
@@ -3,6 +3,29 @@
   tags:
     - galah
   files:
+    - path: output/galah/test-dereplicated/GCA_002688505.1_ASM268850v1_genomic.fna.gz
+      md5sum: 0747c48f6693a4fb03c7164c2f472326
+    - path: output/galah/test-dereplicated/GCF_004296495.1_ASM429649v1_genomic.fna.gz
+      md5sum: a8e9bac598df938f25e09418ff7214dd
     - path: output/galah/test-dereplicated_bins.tsv
       md5sum: d2f8a621bfa5794467f4fdd759e2bce7
     - path: output/galah/versions.yml
+
+- name: galah test_galah_genomeinfo
+  command: nextflow run ./tests/modules/nf-core/galah -entry test_galah_genomeinfo -c ./tests/config/nextflow.config -c ./tests/modules/nf-core/galah/nextflow.config
+  tags:
+    - galah
+  files:
+    - path: output/bioawk/genomeinfo.gz
+      md5sum: 7c7c29f7d08468cb67dfb5635aa144db
+    - path: output/bioawk/versions.yml
+    - path: output/galah/test-dereplicated/GCA_002688505.1_ASM268850v1_genomic.fna.gz
+      md5sum: 0747c48f6693a4fb03c7164c2f472326
+    - path: output/galah/test-dereplicated/GCF_004296495.1_ASM429649v1_genomic.fna.gz
+      md5sum: a8e9bac598df938f25e09418ff7214dd
+    - path: output/galah/test-dereplicated_bins.tsv
+      md5sum: d2f8a621bfa5794467f4fdd759e2bce7
+    - path: output/galah/versions.yml
+    - path: output/gunzip/genomeinfo
+      md5sum: f73b9131ab91ddb754725b94e5085955
+    - path: output/gunzip/versions.yml

From a59006a07d745c89857947904b705dd7eb3f9978 Mon Sep 17 00:00:00 2001
From: prototaxites <jimdownie@fastmail.org>
Date: Tue, 25 Jul 2023 18:10:49 +0000
Subject: [PATCH 08/19] Update error message.

---
 modules/nf-core/galah/main.nf | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/nf-core/galah/main.nf b/modules/nf-core/galah/main.nf
index 6effe726fb4..af95804bd3f 100644
--- a/modules/nf-core/galah/main.nf
+++ b/modules/nf-core/galah/main.nf
@@ -24,7 +24,7 @@ process GALAH {
     def args = task.ext.args ?: ''
     prefix = task.ext.prefix ?: "${meta.id}"
     def qc_input = qc_table ? ((qc_format == "checkm") ? "--checkm-tab-table ${qc_table}" : "--genome-info ${qc_table}") : ""
-    if( qc_table && !(qc_format in ["checkm", "genome_info"]) ) error "Invalid qc_format supplied!"
+    if( qc_table && !(qc_format in ["checkm", "genome_info"]) ) error "Invalid qc_format supplied! qc_format should be either 'checkm' or 'genome_info'."
     """
     mkdir ${prefix}-dereplicated
 

From cb6916c60fba04ec291a3181a10e3e3e6fce15da Mon Sep 17 00:00:00 2001
From: Jim Downie <19718667+prototaxites@users.noreply.github.com>
Date: Thu, 27 Jul 2023 08:26:55 +0100
Subject: [PATCH 09/19] Update main.nf

---
 modules/nf-core/galah/main.nf | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/modules/nf-core/galah/main.nf b/modules/nf-core/galah/main.nf
index af95804bd3f..0eb4a59bc70 100644
--- a/modules/nf-core/galah/main.nf
+++ b/modules/nf-core/galah/main.nf
@@ -2,10 +2,10 @@ process GALAH {
     tag "$meta.id"
     label 'process_medium'
 
-    conda "bioconda::galah=0.3.1 bioconda::dashing=0.4.0 bioconda::fastani=1.31"
+    conda "bioconda::galah=0.3.1"
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'https://depot.galaxyproject.org/singularity/galah:0.3.1--hec16e2b_1':
-        'biocontainers/galah:0.3.1--hec16e2b_1' }"
+        'https://depot.galaxyproject.org/singularity/galah%3A0.3.1--h031d066_3':
+        'biocontainers/galah:0.3.1--h031d066_3' }"
 
     input:
     tuple val(meta), path(bins)

From 798e7550d624f9bff0aa8dc6401a67f64717c7b2 Mon Sep 17 00:00:00 2001
From: Jim Downie <19718667+prototaxites@users.noreply.github.com>
Date: Thu, 27 Jul 2023 08:48:54 +0100
Subject: [PATCH 10/19] Update test.yml

---
 tests/modules/nf-core/galah/test.yml | 2 --
 1 file changed, 2 deletions(-)

diff --git a/tests/modules/nf-core/galah/test.yml b/tests/modules/nf-core/galah/test.yml
index 9eec3cc7eb7..9075894f138 100644
--- a/tests/modules/nf-core/galah/test.yml
+++ b/tests/modules/nf-core/galah/test.yml
@@ -16,8 +16,6 @@
   tags:
     - galah
   files:
-    - path: output/bioawk/genomeinfo.gz
-      md5sum: 7c7c29f7d08468cb67dfb5635aa144db
     - path: output/bioawk/versions.yml
     - path: output/galah/test-dereplicated/GCA_002688505.1_ASM268850v1_genomic.fna.gz
       md5sum: 0747c48f6693a4fb03c7164c2f472326

From 6121ca06ff2a0c0fecdafe0143515e976369fbf1 Mon Sep 17 00:00:00 2001
From: Jim Downie <19718667+prototaxites@users.noreply.github.com>
Date: Thu, 3 Aug 2023 14:24:20 +0100
Subject: [PATCH 11/19] Update main.nf

---
 modules/nf-core/galah/main.nf | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/modules/nf-core/galah/main.nf b/modules/nf-core/galah/main.nf
index 0eb4a59bc70..7edd2cb7c27 100644
--- a/modules/nf-core/galah/main.nf
+++ b/modules/nf-core/galah/main.nf
@@ -23,8 +23,10 @@ process GALAH {
     script:
     def args = task.ext.args ?: ''
     prefix = task.ext.prefix ?: "${meta.id}"
-    def qc_input = qc_table ? ((qc_format == "checkm") ? "--checkm-tab-table ${qc_table}" : "--genome-info ${qc_table}") : ""
-    if( qc_table && !(qc_format in ["checkm", "genome_info"]) ) error "Invalid qc_format supplied! qc_format should be either 'checkm' or 'genome_info'."
+    def qc_args = (qc_format == "checkm") ? "--checkm-tab-table ${qc_table}" : "--genome-info ${qc_table}")
+    def qc_input = qc_table ? qc_args : ""
+    def valid_qc_format = qc_format in ["checkm", "genome_info"])
+    if( qc_table && !valid_qc_format ) error "Invalid qc_format supplied! qc_format should be either 'checkm' or 'genome_info'."
     """
     mkdir ${prefix}-dereplicated
 

From 430c0cbdb37f89b2b898c677371d64d8440c1e5a Mon Sep 17 00:00:00 2001
From: Jim Downie <19718667+prototaxites@users.noreply.github.com>
Date: Fri, 4 Aug 2023 10:02:13 +0100
Subject: [PATCH 12/19] Update modules/nf-core/galah/main.nf

Co-authored-by: Adam Talbot <12817534+adamrtalbot@users.noreply.github.com>
---
 modules/nf-core/galah/main.nf | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/modules/nf-core/galah/main.nf b/modules/nf-core/galah/main.nf
index 7edd2cb7c27..7038b7b4c4a 100644
--- a/modules/nf-core/galah/main.nf
+++ b/modules/nf-core/galah/main.nf
@@ -26,7 +26,9 @@ process GALAH {
     def qc_args = (qc_format == "checkm") ? "--checkm-tab-table ${qc_table}" : "--genome-info ${qc_table}")
     def qc_input = qc_table ? qc_args : ""
     def valid_qc_format = qc_format in ["checkm", "genome_info"])
-    if( qc_table && !valid_qc_format ) error "Invalid qc_format supplied! qc_format should be either 'checkm' or 'genome_info'."
+    if( qc_table && !valid_qc_format ) {
+        error "Invalid qc_format supplied! qc_format should be either 'checkm' or 'genome_info'."
+    }
     """
     mkdir ${prefix}-dereplicated
 

From 08fed34ec0ece30c6506f02e791c5fc060756743 Mon Sep 17 00:00:00 2001
From: Jim Downie <19718667+prototaxites@users.noreply.github.com>
Date: Fri, 4 Aug 2023 10:33:13 +0100
Subject: [PATCH 13/19] Update main.nf

---
 modules/nf-core/galah/main.nf | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/nf-core/galah/main.nf b/modules/nf-core/galah/main.nf
index 7038b7b4c4a..215f624c830 100644
--- a/modules/nf-core/galah/main.nf
+++ b/modules/nf-core/galah/main.nf
@@ -25,7 +25,7 @@ process GALAH {
     prefix = task.ext.prefix ?: "${meta.id}"
     def qc_args = (qc_format == "checkm") ? "--checkm-tab-table ${qc_table}" : "--genome-info ${qc_table}")
     def qc_input = qc_table ? qc_args : ""
-    def valid_qc_format = qc_format in ["checkm", "genome_info"])
+    def valid_qc_format = qc_format in ["checkm", "genome_info"]
     if( qc_table && !valid_qc_format ) {
         error "Invalid qc_format supplied! qc_format should be either 'checkm' or 'genome_info'."
     }

From 390fe417201546280320450802d7c78a9b4b0aec Mon Sep 17 00:00:00 2001
From: Jim Downie <19718667+prototaxites@users.noreply.github.com>
Date: Fri, 4 Aug 2023 10:33:32 +0100
Subject: [PATCH 14/19] Update modules/nf-core/galah/main.nf

Co-authored-by: Adam Talbot <12817534+adamrtalbot@users.noreply.github.com>
---
 modules/nf-core/galah/main.nf | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/modules/nf-core/galah/main.nf b/modules/nf-core/galah/main.nf
index 215f624c830..cef43a42c10 100644
--- a/modules/nf-core/galah/main.nf
+++ b/modules/nf-core/galah/main.nf
@@ -8,9 +8,7 @@ process GALAH {
         'biocontainers/galah:0.3.1--h031d066_3' }"
 
     input:
-    tuple val(meta), path(bins)
-    path(qc_table)
-    val(qc_format)
+    tuple val(meta), path(bins), path(qc_table), val(qc_format)
 
     output:
     tuple val(meta), path("*.tsv")                    , emit: tsv

From ed6c688e9a00185fb76c9ac5ec97936a2857052f Mon Sep 17 00:00:00 2001
From: Jim Downie <19718667+prototaxites@users.noreply.github.com>
Date: Fri, 4 Aug 2023 12:26:11 +0100
Subject: [PATCH 15/19] Update main.nf

---
 modules/nf-core/galah/main.nf | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/nf-core/galah/main.nf b/modules/nf-core/galah/main.nf
index cef43a42c10..19721a5d684 100644
--- a/modules/nf-core/galah/main.nf
+++ b/modules/nf-core/galah/main.nf
@@ -21,7 +21,7 @@ process GALAH {
     script:
     def args = task.ext.args ?: ''
     prefix = task.ext.prefix ?: "${meta.id}"
-    def qc_args = (qc_format == "checkm") ? "--checkm-tab-table ${qc_table}" : "--genome-info ${qc_table}")
+    def qc_args = (qc_format == "checkm") ? "--checkm-tab-table ${qc_table}" : "--genome-info ${qc_table}"
     def qc_input = qc_table ? qc_args : ""
     def valid_qc_format = qc_format in ["checkm", "genome_info"]
     if( qc_table && !valid_qc_format ) {

From daaff42ba7a25c8ea98837a5b1782da6b989f53d Mon Sep 17 00:00:00 2001
From: Jim Downie <19718667+prototaxites@users.noreply.github.com>
Date: Mon, 7 Aug 2023 10:05:34 +0100
Subject: [PATCH 16/19] Update test

---
 tests/modules/nf-core/galah/main.nf | 24 +++++++++++++++---------
 1 file changed, 15 insertions(+), 9 deletions(-)

diff --git a/tests/modules/nf-core/galah/main.nf b/tests/modules/nf-core/galah/main.nf
index d260cc67ed0..2839a7f9452 100644
--- a/tests/modules/nf-core/galah/main.nf
+++ b/tests/modules/nf-core/galah/main.nf
@@ -9,21 +9,17 @@ workflow test_galah {
     input = [
         [ id:'test' ], // meta map
         [file("https://github.com/nf-core/test-datasets/raw/magmap/testdata/GCA_002688505.1_ASM268850v1_genomic.fna.gz", checkIfExists: true),
-        file("https://github.com/nf-core/test-datasets/raw/magmap/testdata/GCF_004296495.1_ASM429649v1_genomic.fna.gz", checkIfExists: true)]
+        file("https://github.com/nf-core/test-datasets/raw/magmap/testdata/GCF_004296495.1_ASM429649v1_genomic.fna.gz", checkIfExists: true)],
+        [], 
+        []
     ]
 
-    GALAH ( input, [], [] )
+    GALAH ( input )
 
 }
 
 workflow test_galah_genomeinfo {
 
-    input = [
-        [ id:'test' ], // meta map
-        [file("https://github.com/nf-core/test-datasets/raw/magmap/testdata/GCA_002688505.1_ASM268850v1_genomic.fna.gz", checkIfExists: true),
-        file("https://github.com/nf-core/test-datasets/raw/magmap/testdata/GCF_004296495.1_ASM429649v1_genomic.fna.gz", checkIfExists: true)]
-    ]
-
     genomeinfo = [
         [ id: 'genomeinfo' ],
         file("https://raw.githubusercontent.com/nf-core/test-datasets/magmap/testdata/checkm.lineage_wf.qa_2.tsv", checkIfExists: true)
@@ -36,6 +32,16 @@ workflow test_galah_genomeinfo {
     ch_genomeinfo = GUNZIP.out.gunzip
         .map { meta, tsv -> [tsv] }
 
-    GALAH ( input, ch_genomeinfo, "genome_info")
+    input = [
+        [ id:'test' ], // meta map
+        [file("https://github.com/nf-core/test-datasets/raw/magmap/testdata/GCA_002688505.1_ASM268850v1_genomic.fna.gz", checkIfExists: true),
+        file("https://github.com/nf-core/test-datasets/raw/magmap/testdata/GCF_004296495.1_ASM429649v1_genomic.fna.gz", checkIfExists: true)],
+        ch_genomeinfo,
+        "genome_info"
+    ]
+
+
+
+    GALAH ( input )
 
 }

From cc26bc3de28f89333858aee0f19aa34378471901 Mon Sep 17 00:00:00 2001
From: Jim Downie <19718667+prototaxites@users.noreply.github.com>
Date: Mon, 7 Aug 2023 10:18:17 +0100
Subject: [PATCH 17/19] Update main.nf

---
 tests/modules/nf-core/galah/main.nf | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/tests/modules/nf-core/galah/main.nf b/tests/modules/nf-core/galah/main.nf
index 2839a7f9452..a5ebdd68952 100644
--- a/tests/modules/nf-core/galah/main.nf
+++ b/tests/modules/nf-core/galah/main.nf
@@ -35,12 +35,12 @@ workflow test_galah_genomeinfo {
     input = [
         [ id:'test' ], // meta map
         [file("https://github.com/nf-core/test-datasets/raw/magmap/testdata/GCA_002688505.1_ASM268850v1_genomic.fna.gz", checkIfExists: true),
-        file("https://github.com/nf-core/test-datasets/raw/magmap/testdata/GCF_004296495.1_ASM429649v1_genomic.fna.gz", checkIfExists: true)],
-        ch_genomeinfo,
-        "genome_info"
+        file("https://github.com/nf-core/test-datasets/raw/magmap/testdata/GCF_004296495.1_ASM429649v1_genomic.fna.gz", checkIfExists: true)]
     ]
-
-
+    .combine(ch_genomeinfo)
+    .map {meta, bins, qc ->
+            [ meta, bins, qc, "genome_info" ]
+    }
 
     GALAH ( input )
 

From 52dcde22da278ba480585ee0482170ab6a8e194a Mon Sep 17 00:00:00 2001
From: Jim Downie <19718667+prototaxites@users.noreply.github.com>
Date: Mon, 7 Aug 2023 10:33:46 +0100
Subject: [PATCH 18/19] Update main.nf

---
 tests/modules/nf-core/galah/main.nf | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/tests/modules/nf-core/galah/main.nf b/tests/modules/nf-core/galah/main.nf
index a5ebdd68952..35ac44ec506 100644
--- a/tests/modules/nf-core/galah/main.nf
+++ b/tests/modules/nf-core/galah/main.nf
@@ -20,10 +20,10 @@ workflow test_galah {
 
 workflow test_galah_genomeinfo {
 
-    genomeinfo = [
-        [ id: 'genomeinfo' ],
-        file("https://raw.githubusercontent.com/nf-core/test-datasets/magmap/testdata/checkm.lineage_wf.qa_2.tsv", checkIfExists: true)
-    ]
+    genomeinfo = Channel.fromPath("https://raw.githubusercontent.com/nf-core/test-datasets/magmap/testdata/checkm.lineage_wf.qa_2.tsv", checkIfExists: true)
+        .map { file ->
+            [ [id: "genomeinfo], file ]
+    }
 
     BIOAWK_GENOMEINFO(genomeinfo)
 

From 673e2f04740072c7f8d0eb48cc40f700be1dd9de Mon Sep 17 00:00:00 2001
From: prototaxites <jimdownie@fastmail.org>
Date: Mon, 7 Aug 2023 10:21:36 +0000
Subject: [PATCH 19/19] fix tests

---
 tests/modules/nf-core/galah/main.nf         | 18 ++++++++++--------
 tests/modules/nf-core/galah/nextflow.config |  1 +
 tests/modules/nf-core/galah/test.yml        |  3 ++-
 3 files changed, 13 insertions(+), 9 deletions(-)

diff --git a/tests/modules/nf-core/galah/main.nf b/tests/modules/nf-core/galah/main.nf
index 35ac44ec506..dba70de9169 100644
--- a/tests/modules/nf-core/galah/main.nf
+++ b/tests/modules/nf-core/galah/main.nf
@@ -10,7 +10,7 @@ workflow test_galah {
         [ id:'test' ], // meta map
         [file("https://github.com/nf-core/test-datasets/raw/magmap/testdata/GCA_002688505.1_ASM268850v1_genomic.fna.gz", checkIfExists: true),
         file("https://github.com/nf-core/test-datasets/raw/magmap/testdata/GCF_004296495.1_ASM429649v1_genomic.fna.gz", checkIfExists: true)],
-        [], 
+        [],
         []
     ]
 
@@ -22,21 +22,23 @@ workflow test_galah_genomeinfo {
 
     genomeinfo = Channel.fromPath("https://raw.githubusercontent.com/nf-core/test-datasets/magmap/testdata/checkm.lineage_wf.qa_2.tsv", checkIfExists: true)
         .map { file ->
-            [ [id: "genomeinfo], file ]
+            [ [id: "genomeinfo"], file ]
     }
 
     BIOAWK_GENOMEINFO(genomeinfo)
 
     GUNZIP(BIOAWK_GENOMEINFO.out.output)
-    
+
     ch_genomeinfo = GUNZIP.out.gunzip
         .map { meta, tsv -> [tsv] }
 
-    input = [
-        [ id:'test' ], // meta map
-        [file("https://github.com/nf-core/test-datasets/raw/magmap/testdata/GCA_002688505.1_ASM268850v1_genomic.fna.gz", checkIfExists: true),
-        file("https://github.com/nf-core/test-datasets/raw/magmap/testdata/GCF_004296495.1_ASM429649v1_genomic.fna.gz", checkIfExists: true)]
-    ]
+    input = Channel.of(
+        [
+            [ id:'test' ], // meta map
+            [file("https://github.com/nf-core/test-datasets/raw/magmap/testdata/GCA_002688505.1_ASM268850v1_genomic.fna.gz", checkIfExists: true),
+            file("https://github.com/nf-core/test-datasets/raw/magmap/testdata/GCF_004296495.1_ASM429649v1_genomic.fna.gz", checkIfExists: true)]
+        ]
+        )
     .combine(ch_genomeinfo)
     .map {meta, bins, qc ->
             [ meta, bins, qc, "genome_info" ]
diff --git a/tests/modules/nf-core/galah/nextflow.config b/tests/modules/nf-core/galah/nextflow.config
index 8eacddfbc81..8cae6f9910f 100644
--- a/tests/modules/nf-core/galah/nextflow.config
+++ b/tests/modules/nf-core/galah/nextflow.config
@@ -7,6 +7,7 @@ process {
     // the file name minus the last extension
     withName: BIOAWK_GENOMEINFO {
         ext.args = '\'BEGIN {{FS="\t"; OFS=","}} NR==1 {print "genome","completeness","contamination"} NR>1 {print $1".fna",$6, $7}\''
+        ext.prefix = "genome_info.tsv"
     }
 
 }
diff --git a/tests/modules/nf-core/galah/test.yml b/tests/modules/nf-core/galah/test.yml
index 9075894f138..47f9e900e96 100644
--- a/tests/modules/nf-core/galah/test.yml
+++ b/tests/modules/nf-core/galah/test.yml
@@ -16,6 +16,7 @@
   tags:
     - galah
   files:
+    - path: output/bioawk/genome_info.tsv.gz
     - path: output/bioawk/versions.yml
     - path: output/galah/test-dereplicated/GCA_002688505.1_ASM268850v1_genomic.fna.gz
       md5sum: 0747c48f6693a4fb03c7164c2f472326
@@ -24,6 +25,6 @@
     - path: output/galah/test-dereplicated_bins.tsv
       md5sum: d2f8a621bfa5794467f4fdd759e2bce7
     - path: output/galah/versions.yml
-    - path: output/gunzip/genomeinfo
+    - path: output/gunzip/genome_info.tsv
       md5sum: f73b9131ab91ddb754725b94e5085955
     - path: output/gunzip/versions.yml