Seqtk sample

CHANGELOG.md

@@ -50,6 +50,9 @@
 
 * `falco`: A C++ drop-in replacement of FastQC to assess the quality of sequence read data (PR #43).
 
+* `seqtk/seqtk_sample`: Sample sequences from FASTA/Q(.gz) files 
+to FASTA/Q (PR #68).
+
 
 ## MAJOR CHANGES
 

src/seqtk/seqtk_sample/config.vsh.yaml

@@ -0,0 +1,54 @@
+name: seqtk_sample
+namespace: seqtk
+description: Subsamples sequences from FASTA/Q files.
+keywords: [tag1, tag2]
+links:
+  repository: https://github.com/lh3/seqtk/tree/v1.4
+license: MIT
+
+argument_groups:
+  - name: Inputs
+    arguments:
+      - name: --input
+        type: file
+        description: The input FASTA/Q file.
+        required: true
+
+  - name: Outputs
+    arguments:
+      - name: --output
+        type: file
+        description: The output FASTA/Q file.
+        required: true
+        direction: output
+
+  - name: Options
+    arguments:
+      - name: --seed
+        type: integer
+        description: Seed for random generator.
+        default: 42
+      - name: --fraction_number
+        type: double
+        description: Fraction or number of sequences to sample.
+        default: 0.1
+      - name: --two_pass_mode
+        type: boolean
+        description: twice as slow but with much reduced memory
+        default: false
+
+resources:
+  - type: bash_script
+    path: script.sh
+test_resources:
+  - type: bash_script
+    path: test.sh
+  - type: file
+    path: test_data
+
+engines:
+  - type: docker
+    image: quay.io/biocontainers/seqtk:1.4--he4a0461_2
+runners:
+  - type: executable
+  - type: nextflow

src/seqtk/seqtk_sample/help.txt

@@ -0,0 +1,7 @@
+```
+seqtk_sample
+```
+Usage:   seqtk sample [-2] [-s seed=11] <in.fa> <frac>|<number> > <out.fa>
+
+Options: -s INT       RNG seed [11]
+         -2           2-pass mode: twice as slow but with much reduced memory

src/seqtk/seqtk_sample/script.sh

@@ -0,0 +1,11 @@
+#!/bin/bash
+
+## VIASH START
+## VIASH END
+
+seqtk sample \
+    ${par_2_pass_mode:+-2} \
+    ${par_seed:+-s "$par_seed"} \
+    "$par_input" \
+    "$par_fraction_number" \
+    > "$par_output"

src/seqtk/seqtk_sample/test.sh

@@ -0,0 +1,90 @@
+#!/bin/bash
+
+set -e
+
+## VIASH START
+meta_executable="target/executable/seqtk/seqtk_sample"
+meta_resources_dir="src/seqtk/seqtk_sample"
+## VIASH END
+
+#########################################################################################
+mkdir seqtk_sample_se
+cd seqtk_sample_se
+
+echo "> Run seqtk_sample on fastq SE"
+"$meta_executable" \
+  --input "$meta_resources_dir/test_data/reads/a.fastq" \
+  --seed 42 \
+  --fraction_number 3 \
+  --output "sampled.fastq"
+
+echo ">> Check if output exists"
+if [ ! -f "sampled.fastq" ]; then
+    echo ">> sampled.fastq.gz does not exist"
+    exit 1
+fi
+
+#########################################################################################
+cd ..
+mkdir seqtk_sample_pe_number
+cd seqtk_sample_pe_number
+
+echo ">> Run seqtk_sample on fastq.gz PE with number of reads"
+"$meta_executable" \
+  --input "$meta_resources_dir/test_data/reads/a.1.fastq.gz" \
+  --seed 42 \
+  --fraction_number 3 \
+  --output "sampled_1.fastq"
+
+"$meta_executable" \
+  --input "$meta_resources_dir/test_data/reads/a.2.fastq.gz" \
+  --seed 42 \
+  --fraction_number 3 \
+  --output "sampled_2.fastq"
+
+echo ">> Check if output exists"
+if [ ! -f "sampled_1.fastq" ] || [ ! -f "sampled_2.fastq" ]; then
+    echo ">> One or both output files do not exist"
+    exit 1
+fi
+
+echo ">> Compare reads"
+# Extract headers
+headers1=$(grep '^@' sampled_1.fastq | sed -e's/ 1$//' | sort)
+headers2=$(grep '^@' sampled_2.fastq | sed -e 's/ 2$//' | sort)
+
+# Compare headers
+diff <(echo "$headers1") <(echo "$headers2") || { echo "Mismatch detected"; exit 1; }
+
+#########################################################################################
+cd ..
+mkdir seqtk_sample_pe_fraction
+cd seqtk_sample_pe_fraction
+
+echo ">> Run seqtk_sample on fastq.gz PE with fraction of reads"
+"$meta_executable" \
+  --input "$meta_resources_dir/test_data/reads/a.1.fastq.gz" \
+  --seed 42 \
+  --fraction_number 0.5 \
+  --output "sampled_1.fastq"
+
+"$meta_executable" \
+  --input "$meta_resources_dir/test_data/reads/a.2.fastq.gz" \
+  --seed 42 \
+  --fraction_number 0.5 \
+  --output "sampled_2.fastq"
+
+echo ">> Check if output exists"
+if [ ! -f "sampled_1.fastq" ] || [ ! -f "sampled_2.fastq" ]; then
+    echo ">> One or both output files do not exist"
+    exit 1
+fi
+
+echo ">> Compare reads"
+# Extract headers
+headers1=$(grep '^@' sampled_1.fastq | sed -e's/ 1$//' | sort)
+headers2=$(grep '^@' sampled_2.fastq | sed -e 's/ 2$//' | sort)
+
+# Compare headers
+diff <(echo "$headers1") <(echo "$headers2") || { echo "Mismatch detected"; exit 1; }
+

src/seqtk/seqtk_sample/test_data/reads/a.fastq

@@ -0,0 +1,4 @@
+@1
+ACGGCAT
++
+!!!!!!!

src/seqtk/seqtk_sample/test_data/reads/id.list

@@ -0,0 +1 @@
+1

src/seqtk/seqtk_sample/test_data/script.sh

@@ -0,0 +1,9 @@
+# clone repo
+if [ ! -d /tmp/snakemake-wrappers ]; then
+  git clone --depth 1 --single-branch --branch master https://github.com/snakemake/snakemake-wrappers /tmp/snakemake-wrappers
+fi
+
+# copy test data
+cp -r /tmp/snakemake-wrappers/bio/seqtk/test/* src/seqtk/seqtk_sample/test_data
+
+rm src/seqtk/seqtk_sample/test_data/Snakefile