From 599f37078dbe669106a867845da0caddfa259908 Mon Sep 17 00:00:00 2001
From: William Rowell <williamrowell@users.noreply.github.com>
Date: Wed, 26 Jan 2022 06:40:31 -0800
Subject: [PATCH] feat: added wrapper for seqtk seq subcommand (#202)

* added wrapper for seqtk seq subcommand

* added test and linted
---
 bio/seqtk/seq/environment.yaml |  5 +++++
 bio/seqtk/seq/meta.yaml        | 10 ++++++++++
 bio/seqtk/seq/test/Snakefile   | 11 +++++++++++
 bio/seqtk/seq/test/a.fastq     |  4 ++++
 bio/seqtk/seq/wrapper.py       | 15 +++++++++++++++
 test.py                        |  8 ++++++++
 6 files changed, 53 insertions(+)
 create mode 100644 bio/seqtk/seq/environment.yaml
 create mode 100644 bio/seqtk/seq/meta.yaml
 create mode 100644 bio/seqtk/seq/test/Snakefile
 create mode 100644 bio/seqtk/seq/test/a.fastq
 create mode 100644 bio/seqtk/seq/wrapper.py

diff --git a/bio/seqtk/seq/environment.yaml b/bio/seqtk/seq/environment.yaml
new file mode 100644
index 0000000000..b0be2c88b3
--- /dev/null
+++ b/bio/seqtk/seq/environment.yaml
@@ -0,0 +1,5 @@
+channels:
+  - bioconda
+  - conda-forge
+dependencies:
+  - seqtk ==1.3
diff --git a/bio/seqtk/seq/meta.yaml b/bio/seqtk/seq/meta.yaml
new file mode 100644
index 0000000000..6c621f27b1
--- /dev/null
+++ b/bio/seqtk/seq/meta.yaml
@@ -0,0 +1,10 @@
+name: seqtk-seq
+description: |
+  Common transformations of FASTA/Q using seqtk
+url: https://github.com/lh3/seqtk
+authors:
+  - William Rowell
+input:
+  - fastn file (can be gzip compressed)
+output:
+  - fastn file (gzip compressed)
diff --git a/bio/seqtk/seq/test/Snakefile b/bio/seqtk/seq/test/Snakefile
new file mode 100644
index 0000000000..0bf469a6ce
--- /dev/null
+++ b/bio/seqtk/seq/test/Snakefile
@@ -0,0 +1,11 @@
+rule seqtk_seq_fastq_to_fasta:
+    input:
+        "{prefix}.fastq",
+    output:
+        "{prefix}.fasta",
+    log:
+        "{prefix}.log",
+    params:
+        extra="-A",
+    wrapper:
+        "master/bio/seqtk/seq"
diff --git a/bio/seqtk/seq/test/a.fastq b/bio/seqtk/seq/test/a.fastq
new file mode 100644
index 0000000000..42735560ae
--- /dev/null
+++ b/bio/seqtk/seq/test/a.fastq
@@ -0,0 +1,4 @@
+@1
+ACGGCAT
++
+!!!!!!!
diff --git a/bio/seqtk/seq/wrapper.py b/bio/seqtk/seq/wrapper.py
new file mode 100644
index 0000000000..9ecccfaf2d
--- /dev/null
+++ b/bio/seqtk/seq/wrapper.py
@@ -0,0 +1,15 @@
+"""Snakemake wrapper seqtk seq subcommand"""
+
+__author__ = "William Rowell"
+__copyright__ = "Copyright 2020, William Rowell"
+__email__ = "wrowell@pacb.com"
+__license__ = "MIT"
+
+
+from snakemake.shell import shell
+
+
+extra = snakemake.params.get("extra", "")
+log = snakemake.log_fmt_shell(stdout=False, stderr=True)
+
+shell("(seqtk seq {extra} {snakemake.input} > {snakemake.output}) {log}")
diff --git a/test.py b/test.py
index 2c4fa2bd14..bd0fa12950 100644
--- a/test.py
+++ b/test.py
@@ -605,6 +605,14 @@ def test_shovill():
     )
 
 
+@skip_if_not_modified
+def test_seqtk_seq():
+    run(
+        "bio/seqtk/seq",
+        ["snakemake", "--cores", "1", "--use-conda", "-F", "a.fasta"],
+    )
+
+
 @skip_if_not_modified
 def test_seqtk_mergepe():
     run(