Fq subsample

CHANGELOG.md

@@ -131,6 +131,8 @@
     - `bedtools_getfasta`: extract sequences from a FASTA file for each of the
                            intervals defined in a BED/GFF/VCF file (PR #59).
 
+*  `fq_subsample`: Sample a subset of records from single or paired FASTQ files (PR #147).
+
 ## MINOR CHANGES
 
 * Uniformize component metadata (PR #23).

src/fq_subsample/config.vsh.yaml

@@ -0,0 +1,73 @@
+name: fq_subsample
+description: fq subsample outputs a subset of records from single or paired FASTQ files.
+keywords: [fastq, subsample, subset]
+links:
+  homepage: https://github.com/stjude-rust-labs/fq/blob/master/README.md
+  documentation: https://github.com/stjude-rust-labs/fq/blob/master/README.md
+  repository: https://github.com/stjude-rust-labs/fq
+license: MIT
+
+argument_groups: 
+- name: "Input"
+  arguments: 
+  - name: "--input_1"
+    type: file
+    required: true
+    description: First input fastq file to subsample. Accepts both raw and gzipped FASTQ inputs.
+  - name: "--input_2"
+    type: file
+    description: Second input fastq files to subsample. Accepts both raw and gzipped FASTQ inputs.
+
+- name: "Output"
+  arguments: 
+  - name: "--output_1"
+    type: file
+    direction: output
+    description: Sampled read 1 fastq files. Output will be gzipped if ends in `.gz`.
+  - name: "--output_2"
+    type: file
+    direction: output
+    description: Sampled read 2 fastq files. Output will be gzipped if ends in `.gz`.
+
+- name: "Options"
+  arguments: 
+  - name: "--probability"
+    type: double
+    description: The probability a record is kept, as a percentage (0.0, 1.0). Cannot be used with `record-count`
+  - name: "--record_count"
+    type: integer
+    description: The exact number of records to keep. Cannot be used with `probability`
+  - name: "--seed"
+    type: integer
+    description: Seed to use for the random number generator
+
+resources:
+  - type: bash_script
+    path: script.sh
+
+test_resources:
+  - type: bash_script
+    path: test.sh
+  - path: test_data
+
+engines:
+  - type: docker
+    image: ubuntu:22.04
+    setup: 
+      - type: docker
+        env: 
+            - TZ=Europe/Brussels
+        run: | 
+          ln -snf /usr/share/zoneinfo/$TZ /etc/localtime && echo $TZ > /etc/timezone && \
+          apt-get update && \
+          apt-get install -y --no-install-recommends build-essential git-all curl && \
+          curl https://sh.rustup.rs -sSf | sh -s -- -y && \
+          . "$HOME/.cargo/env" && \
+          git clone --depth 1 --branch v0.12.0 https://github.com/stjude-rust-labs/fq.git && \
+          mv fq /usr/local/ && cd /usr/local/fq && \
+          cargo install --locked --path . && \
+          mv /usr/local/fq/target/release/fq /usr/local/bin/
+
+runners:
+  - type: executable
+  - type: nextflow

src/fq_subsample/help.txt

@@ -0,0 +1,20 @@
+```
+fq subsample -h
+```
+
+Outputs a subset of records
+
+Usage: fq subsample [OPTIONS] --r1-dst <R1_DST> <--probability <PROBABILITY>|--record-count <RECORD_COUNT>> <R1_SRC> [R2_SRC]
+
+Arguments:
+  <R1_SRC>  Read 1 source. Accepts both raw and gzipped FASTQ inputs
+  [R2_SRC]  Read 2 source. Accepts both raw and gzipped FASTQ inputs
+
+Options:
+  -p, --probability <PROBABILITY>    The probability a record is kept, as a percentage (0.0, 1.0). Cannot be used with `record-count`
+  -n, --record-count <RECORD_COUNT>  The exact number of records to keep. Cannot be used with `probability`
+  -s, --seed <SEED>                  Seed to use for the random number generator
+      --r1-dst <R1_DST>              Read 1 destination. Output will be gzipped if ends in `.gz`
+      --r2-dst <R2_DST>              Read 2 destination. Output will be gzipped if ends in `.gz`
+  -h, --help                         Print help
+  -V, --version   

src/fq_subsample/script.sh

@@ -0,0 +1,26 @@
+#!/bin/bash
+
+## VIASH START
+## VIASH END
+
+set -eo pipefail
+
+
+required_args=("-p" "--probability" "-n" "--record_count")
+
+# exclusive OR for required arguments $par_probability and $par_record_count
+if [[ -n $par_probability && -n $par_record_count ]] || [[ -z $par_probability && -z $par_record_count ]]; then
+    echo "FQ/SUBSAMPLE requires either --probability or --record_count to be specified"
+    exit 1
+fi
+
+
+fq subsample \
+    ${par_output_1:+--r1-dst "${par_output_1}"} \
+    ${par_output_2:+--r2-dst "${par_output_2}"} \
+    ${par_probability:+--probability "${par_probability}"} \
+    ${par_record_count:+--record-count "${par_record_count}"} \
+    ${par_seed:+--seed "${par_seed}"} \
+    ${par_input_1} \
+    ${par_input_2}
+

src/fq_subsample/test.sh

@@ -0,0 +1,36 @@
+#!/bin/bash
+
+echo ">>> Testing $meta_executable"
+
+echo ">>> Testing for paired-end reads"
+"$meta_executable" \
+    --input_1 $meta_resources_dir/test_data/a.3.fastq.gz \
+    --input_2 $meta_resources_dir/test_data/a.4.fastq.gz \
+    --record_count 3 \
+    --seed 1 \
+    --output_1  a.1.subsampled.fastq \
+    --output_2  a.2.subsampled.fastq 
+
+echo ">> Checking if the correct files are present"
+[ ! -f "a.1.subsampled.fastq" ] && echo "Subsampled FASTQ file for read 1 is missing!" && exit 1
+[ $(wc -l < a.1.subsampled.fastq) -ne 12 ] && echo "Subsampled FASTQ file for read 1 does not contain the expected number of records" && exit 1
+[ ! -f "a.2.subsampled.fastq" ] && echo "Subsampled FASTQ file for read 2 is missing" && exit 1
+[ $(wc -l < a.2.subsampled.fastq) -ne 12 ] && echo "Subsampled FASTQ file for read 2 does not contain the expected number of records" && exit 1
+
+rm a.1.subsampled.fastq a.2.subsampled.fastq
+
+echo ">>> Testing for single-end reads"
+"$meta_executable" \
+    --input_1 $meta_resources_dir/test_data/a.3.fastq.gz \
+    --record_count 3 \
+    --seed 1 \
+    --output_1  a.1.subsampled.fastq 
+
+
+echo ">> Checking if the correct files are present"
+[ ! -f "a.1.subsampled.fastq" ] && echo "Subsampled FASTQ file is missing" && exit 1
+[ $(wc -l < a.1.subsampled.fastq) -ne 12 ] && echo "Subsampled FASTQ file does not contain the expected number of records" && exit 1
+
+echo ">>> Tests finished successfully"
+exit 0
+