From 82fdff4fb4ce6cafcc028a7503da835427f35352 Mon Sep 17 00:00:00 2001
From: Jose Espinosa-Carrasco <kadomu@gmail.com>
Date: Thu, 30 Jun 2022 17:23:30 +0200
Subject: [PATCH] Add khmer unique-kmers.py module (#1838)

---
 modules/khmer/uniquekmers/main.nf             | 38 +++++++++++++++++
 modules/khmer/uniquekmers/meta.yml            | 42 +++++++++++++++++++
 tests/config/pytest_modules.yml               |  4 ++
 tests/modules/khmer/uniquekmers/main.nf       | 12 ++++++
 .../modules/khmer/uniquekmers/nextflow.config |  5 +++
 tests/modules/khmer/uniquekmers/test.yml      | 10 +++++
 6 files changed, 111 insertions(+)
 create mode 100644 modules/khmer/uniquekmers/main.nf
 create mode 100644 modules/khmer/uniquekmers/meta.yml
 create mode 100644 tests/modules/khmer/uniquekmers/main.nf
 create mode 100644 tests/modules/khmer/uniquekmers/nextflow.config
 create mode 100644 tests/modules/khmer/uniquekmers/test.yml

diff --git a/modules/khmer/uniquekmers/main.nf b/modules/khmer/uniquekmers/main.nf
new file mode 100644
index 00000000..8ad89a62
--- /dev/null
+++ b/modules/khmer/uniquekmers/main.nf
@@ -0,0 +1,38 @@
+process KHMER_UNIQUEKMERS {
+    tag "$fasta"
+    label 'process_low'
+
+    conda (params.enable_conda ? "bioconda::khmer=3.0.0a3" : null)
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/khmer:3.0.0a3--py37haa7609a_2' :
+        'quay.io/biocontainers/khmer:3.0.0a3--py37haa7609a_2' }"
+
+    input:
+    path fasta
+    val  kmer_size
+
+    output:
+    path "report.txt"  , emit: report
+    path "kmers.txt"   , emit: kmers
+    path "versions.yml", emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+    """
+    unique-kmers.py \\
+        -k $kmer_size \\
+        -R report.txt \\
+        $args \\
+        $fasta
+
+    grep ^number report.txt | sed 's/^.*:.[[:blank:]]//g' > kmers.txt
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        khmer: \$( unique-kmers.py --version 2>&1 | grep ^khmer | sed 's/^khmer //;s/ .*\$//' )
+    END_VERSIONS
+    """
+}
diff --git a/modules/khmer/uniquekmers/meta.yml b/modules/khmer/uniquekmers/meta.yml
new file mode 100644
index 00000000..31405cc1
--- /dev/null
+++ b/modules/khmer/uniquekmers/meta.yml
@@ -0,0 +1,42 @@
+name: "khmer_uniquekmers"
+description:
+keywords:
+  - khmer
+  - k-mer
+  - effective genome size
+
+tools:
+  - "khmer":
+      description: khmer k-mer counting library
+      homepage: https://github.com/dib-lab/khmer
+      documentation: https://khmer.readthedocs.io/en/latest/
+      tool_dev_url: https://github.com/dib-lab/khmer
+      doi: "10.12688/f1000research.6924.1"
+      licence: ["BSD License"]
+
+input:
+  - fasta:
+      type: file
+      description: fasta file
+      pattern: "*.{fa,fasta}"
+  - kmer_size:
+      type: value
+      description: k-mer size to use
+      pattern: "[0-9]+"
+
+output:
+  - report:
+      type: file
+      description: Text file containing unique-kmers.py execution report
+      pattern: "report.txt"
+  - kmers:
+      type: file
+      description: Text file containing number of kmers
+      pattern: "kmers.txt"
+  - versions:
+      type: file
+      description: File containing software versions
+      pattern: "versions.yml"
+
+authors:
+  - "@JoseEspinosa"
diff --git a/tests/config/pytest_modules.yml b/tests/config/pytest_modules.yml
index de4eb424..35a8c897 100644
--- a/tests/config/pytest_modules.yml
+++ b/tests/config/pytest_modules.yml
@@ -1177,6 +1177,10 @@ khmer/normalizebymedian:
   - modules/khmer/normalizebymedian/**
   - tests/modules/khmer/normalizebymedian/**
 
+khmer/uniquekmers:
+  - modules/khmer/uniquekmers/**
+  - tests/modules/khmer/uniquekmers/**
+
 kleborate:
   - modules/kleborate/**
   - tests/modules/kleborate/**
diff --git a/tests/modules/khmer/uniquekmers/main.nf b/tests/modules/khmer/uniquekmers/main.nf
new file mode 100644
index 00000000..cfd6de9f
--- /dev/null
+++ b/tests/modules/khmer/uniquekmers/main.nf
@@ -0,0 +1,12 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { KHMER_UNIQUEKMERS } from '../../../../modules/khmer/uniquekmers/main.nf'
+
+workflow test_khmer_uniquekmers {
+
+    input = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+
+    KHMER_UNIQUEKMERS ( input, 50 )
+}
diff --git a/tests/modules/khmer/uniquekmers/nextflow.config b/tests/modules/khmer/uniquekmers/nextflow.config
new file mode 100644
index 00000000..50f50a7a
--- /dev/null
+++ b/tests/modules/khmer/uniquekmers/nextflow.config
@@ -0,0 +1,5 @@
+process {
+
+    publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
+    
+}
\ No newline at end of file
diff --git a/tests/modules/khmer/uniquekmers/test.yml b/tests/modules/khmer/uniquekmers/test.yml
new file mode 100644
index 00000000..ad4e8a54
--- /dev/null
+++ b/tests/modules/khmer/uniquekmers/test.yml
@@ -0,0 +1,10 @@
+- name: khmer uniquekmers test_khmer_uniquekmers
+  command: nextflow run ./tests/modules/khmer/uniquekmers -entry test_khmer_uniquekmers -c ./tests/config/nextflow.config  -c ./tests/modules/khmer/uniquekmers/nextflow.config
+  tags:
+    - khmer
+    - khmer/uniquekmers
+  files:
+    - path: output/khmer/kmers.txt
+      md5sum: 496ebf23653a01c7a42d743e47c19f65
+    - path: output/khmer/report.txt
+      md5sum: ee489abd3b244dea3640649e1790d55e