From 75606893756ee2e2bf34d650c9973932301fb20a Mon Sep 17 00:00:00 2001
From: "Moritz E. Beber" <midnighter@posteo.net>
Date: Tue, 15 Feb 2022 12:30:28 +0100
Subject: [PATCH] feat: add sourmash sketch (#1287)

Co-authored-by: Jose Espinosa-Carrasco <kadomu@gmail.com>
Co-authored-by: Harshil Patel <drpatelh@users.noreply.github.com>
---
 modules/sourmash/sketch/main.nf               | 35 ++++++++++++++++
 modules/sourmash/sketch/meta.yml              | 42 +++++++++++++++++++
 tests/config/pytest_modules.yml               |  4 ++
 tests/modules/sourmash/sketch/main.nf         | 15 +++++++
 tests/modules/sourmash/sketch/nextflow.config |  5 +++
 tests/modules/sourmash/sketch/test.yml        | 10 +++++
 6 files changed, 111 insertions(+)
 create mode 100644 modules/sourmash/sketch/main.nf
 create mode 100644 modules/sourmash/sketch/meta.yml
 create mode 100644 tests/modules/sourmash/sketch/main.nf
 create mode 100644 tests/modules/sourmash/sketch/nextflow.config
 create mode 100644 tests/modules/sourmash/sketch/test.yml

diff --git a/modules/sourmash/sketch/main.nf b/modules/sourmash/sketch/main.nf
new file mode 100644
index 00000000..e16b605f
--- /dev/null
+++ b/modules/sourmash/sketch/main.nf
@@ -0,0 +1,35 @@
+process SOURMASH_SKETCH {
+    tag "$meta.id"
+    label 'process_low'
+
+    conda (params.enable_conda ? "bioconda::sourmash=4.2.4" : null)
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/sourmash:4.2.4--hdfd78af_0':
+        'quay.io/biocontainers/sourmash:4.2.4--hdfd78af_0' }"
+
+    input:
+    tuple val(meta), path(sequence)
+
+    output:
+    tuple val(meta), path("*.sig"), emit: signatures
+    path "versions.yml"           , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: "dna --param-string 'scaled=1000,k=31'"
+    def prefix = task.ext.prefix ?: "${meta.id}"
+    """
+    sourmash sketch \\
+        $args \\
+        --merge '${prefix}' \\
+        --output '${prefix}.sig' \\
+        $sequence
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        sourmash: \$(echo \$(sourmash --version 2>&1) | sed 's/^sourmash //' )
+    END_VERSIONS
+    """
+}
diff --git a/modules/sourmash/sketch/meta.yml b/modules/sourmash/sketch/meta.yml
new file mode 100644
index 00000000..9fb552bc
--- /dev/null
+++ b/modules/sourmash/sketch/meta.yml
@@ -0,0 +1,42 @@
+name: sourmash_sketch
+description: Create a signature (a hash sketch) of a sequence using sourmash
+keywords:
+  - hash sketch
+  - signature
+tools:
+  - sourmash:
+      description: Compute and compare MinHash signatures for DNA data sets.
+      homepage: https://sourmash.readthedocs.io/
+      documentation: https://sourmash.readthedocs.io/
+      tool_dev_url: https://github.com/dib-lab/sourmash
+      doi: "10.1186/s13059-016-0997-x"
+      licence: ['BSD-3-clause']
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - sequence:
+      type: file
+      description: FastA file containing (genomic) sequence data
+      pattern: "*.{fna,fa,fasta}"
+
+output:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - versions:
+      type: file
+      description: File containing software versions
+      pattern: "versions.yml"
+  - signatures:
+      type: file
+      description: MinHash signature of the given sequence
+      pattern: "*.{sig}"
+
+authors:
+  - "@Midnighter"
diff --git a/tests/config/pytest_modules.yml b/tests/config/pytest_modules.yml
index de061264..52c7f569 100644
--- a/tests/config/pytest_modules.yml
+++ b/tests/config/pytest_modules.yml
@@ -1417,6 +1417,10 @@ snpsites:
   - modules/snpsites/**
   - tests/modules/snpsites/**
 
+sourmash/sketch:
+  - modules/sourmash/sketch/**
+  - tests/modules/sourmash/sketch/**
+
 spades:
   - modules/spades/**
   - tests/modules/spades/**
diff --git a/tests/modules/sourmash/sketch/main.nf b/tests/modules/sourmash/sketch/main.nf
new file mode 100644
index 00000000..c9288c59
--- /dev/null
+++ b/tests/modules/sourmash/sketch/main.nf
@@ -0,0 +1,15 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { SOURMASH_SKETCH } from '../../../../modules/sourmash/sketch/main.nf'
+
+workflow test_sourmash_sketch {
+
+    input = [
+        [ id:'test', single_end:false ], // meta map
+        file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+    ]
+
+    SOURMASH_SKETCH ( input )
+}
diff --git a/tests/modules/sourmash/sketch/nextflow.config b/tests/modules/sourmash/sketch/nextflow.config
new file mode 100644
index 00000000..50f50a7a
--- /dev/null
+++ b/tests/modules/sourmash/sketch/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/sourmash/sketch/test.yml b/tests/modules/sourmash/sketch/test.yml
new file mode 100644
index 00000000..5bd18967
--- /dev/null
+++ b/tests/modules/sourmash/sketch/test.yml
@@ -0,0 +1,10 @@
+- name: sourmash sketch test_sourmash_sketch
+  command: nextflow run tests/modules/sourmash/sketch -entry test_sourmash_sketch -c tests/config/nextflow.config
+  tags:
+    - sourmash
+    - sourmash/sketch
+  files:
+    - path: output/sourmash/test.sig
+      md5sum: 4b0dee307e35fc670cd0d416321e4961
+    - path: output/sourmash/versions.yml
+      md5sum: ec70dabc3e4b3bf11d39c17a99ca4bc6