Merge pull request #1 from ICGC-TCGA-PanCancer/fastqc@0.1.0

[release] fastqc@0.1.0

fastqc/.dockerignore

@@ -0,0 +1,5 @@
+.gitignore
+.nextflow*
+tests
+work
+outdir

fastqc/.gitignore

@@ -0,0 +1,69 @@
+*.py[cod]
+
+# C extensions
+*.so
+
+# Packages
+*.egg
+*.egg-info
+dist
+build
+eggs
+.eggs
+parts
+bin
+var
+sdist
+develop-eggs
+.installed.cfg
+lib
+lib64
+venv*/
+pyvenv*/
+
+# Installer logs
+pip-log.txt
+
+# Unit test / coverage reports
+.coverage
+.tox
+.coverage.*
+nosetests.xml
+coverage.xml
+htmlcov
+
+# Translations
+*.mo
+
+# Mr Developer
+.mr.developer.cfg
+.project
+.pydevproject
+.idea
+*.iml
+*.komodoproject
+
+# Complexity
+output/*.html
+output/*/index.html
+
+# Sphinx
+docs/_build
+
+.DS_Store
+*~
+.*.sw[po]
+.build
+.ve
+.env
+.cache
+.pytest
+.bootstrap
+.appveyor.token
+*.bak
+*.log
+.vscode
+.python-version
+.nextflow*
+work
+outdir

fastqc/Dockerfile

@@ -0,0 +1,9 @@
+FROM pegi3s/fastqc:0.11.9
+
+LABEL org.opencontainers.image.source https://github.com/icgc-tcga-pancancer/demo-tool-pkgs
+
+ENV PATH="/tools:${PATH}"
+
+COPY *.py /tools/
+
+CMD ["/bin/bash"]

fastqc/fastqc.nf

@@ -0,0 +1,54 @@
+#!/usr/bin/env nextflow
+
+/********************************************************************/
+/* this block is auto-generated based on info from pkg.json where   */
+/* changes can be made if needed, do NOT modify this block manually */
+nextflow.enable.dsl = 2
+version = '0.1.0'  // tool version
+
+container = [
+    'ghcr.io': 'ghcr.io/icgc-tcga-pancancer/demo-tool-pkgs.fastqc'
+]
+default_container_registry = 'ghcr.io'
+/********************************************************************/
+
+
+// universal params go here
+params.container_registry = default_container_registry
+params.container_version = ""
+
+params.cpus = 1
+params.mem = 1  // GB
+params.publish_dir = ""  // set to empty string will disable publishDir
+
+
+// tool specific parmas go here, add / change as needed
+params.input_file = ""
+params.output_pattern = "*.html"  // fastqc output html report
+
+
+process fastqc {
+  container "${container[params.container_registry]}:${params.container_version ?: version}"
+  publishDir "${params.publish_dir}/${task.process.replaceAll(':', '_')}", mode: "copy", enabled: "${params.publish_dir ? true : ''}"
+
+  cpus params.cpus
+  memory "${params.mem} GB"
+
+  input:  // input, make update as needed
+    path input_file
+
+  output:  // output, make update as needed
+    path "output_dir/${params.output_pattern}", emit: output
+
+  script:
+    // add and initialize variables here as needed
+
+    """
+    mkdir -p output_dir
+
+    fastqc.py \
+      -i ${input_file} \
+      -o output_dir
+
+    """
+}

fastqc/fastqc.py

@@ -0,0 +1,34 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+
+import os
+import sys
+import argparse
+import subprocess
+
+
+def main():
+    """
+    Python implementation of tool: fastqc
+
+    This is auto-generated Python code, please update as needed!
+    """
+
+    parser = argparse.ArgumentParser(description='Tool: fastqc')
+    parser.add_argument('-i', '--input-file', dest='input_file', type=str,
+                        help='Input file', required=True)
+    parser.add_argument('-o', '--output-dir', dest='output_dir', type=str,
+                        help='Output directory', required=True)
+    args = parser.parse_args()
+
+    if not os.path.isfile(args.input_file):
+        sys.exit('Error: specified input file %s does not exist or is not accessible!' % args.input_file)
+
+    if not os.path.isdir(args.output_dir):
+        sys.exit('Error: specified output dir %s does not exist or is not accessible!' % args.output_dir)
+
+    subprocess.run(f"fastqc -o {args.output_dir} {args.input_file}", shell=True, check=True)
+
+
+if __name__ == "__main__":
+    main()

fastqc/nextflow.config

@@ -0,0 +1,4 @@
+docker {
+    enabled = true
+    runOptions = '-u \$(id -u):\$(id -g)'
+}

fastqc/pkg.json

@@ -0,0 +1,40 @@
+{
+    "name": "fastqc",
+    "version": "0.1.0",
+    "description": "FastQC tool",
+    "main": "fastqc",
+    "scripts": {
+        "test": "wfpm test"
+    },
+    "deprecated": false,
+    "keywords": [
+        "bioinformatics",
+        "seq",
+        "qc metrics"
+    ],
+    "repository": {
+        "type": "git",
+        "url": "https://github.com/icgc-tcga-pancancer/demo-tool-pkgs.git"
+    },
+    "container": {
+        "registries": [
+            {
+                "registry": "ghcr.io",
+                "type": "docker",
+                "org": "icgc-tcga-pancancer",
+                "default": true
+            }
+        ]
+    },
+    "dependencies": [],
+    "devDependencies": [],
+    "contributors": [
+        {
+            "name": "Junjun Zhang",
+            "email": "junjun.ca@gmail.com"
+        }
+    ],
+    "license": "MIT",
+    "bugReport": "https://github.com/icgc-tcga-pancancer/demo-tool-pkgs/issues",
+    "homepage": "https://github.com/icgc-tcga-pancancer/demo-tool-pkgs#readme"
+}

fastqc/tests/checker.nf

@@ -0,0 +1,75 @@
+#!/usr/bin/env nextflow
+
+/********************************************************************/
+/* this block is auto-generated based on info from pkg.json where   */
+/* changes can be made if needed, do NOT modify this block manually */
+nextflow.enable.dsl = 2
+version = '0.1.0'  // tool version
+
+container = [
+    'ghcr.io': 'ghcr.io/icgc-tcga-pancancer/demo-tool-pkgs.fastqc'
+]
+default_container_registry = 'ghcr.io'
+/********************************************************************/
+
+// universal params
+params.container_registry = default_container_registry
+params.container_version = ""
+
+
+// tool specific parmas go here, add / change as needed
+params.input_file = ""
+params.expected_output = ""
+
+include { fastqc } from '../fastqc'
+
+Channel
+  .fromPath(params.input_file, checkIfExists: true)
+  .set { input_file }
+
+
+process file_diff {
+  container "${container[params.container_registry]}:${params.container_version ?: version}"
+
+  input:
+    path output_file
+    path expected_gzip
+
+  output:
+    stdout()
+
+  script:
+    """
+    # remove date field before comparison eg, <div id="header_filename">Tue 19 Jan 2021<br/>test_rg_3.bam</div>
+    # sed -e 's#"header_filename">.*<br/>test_rg_3.bam#"header_filename"><br/>test_rg_3.bam</div>#'
+
+    diff <( cat ${output_file} | sed -e 's#"header_filename">.*<br/>test_rg_3.bam#"header_filename"><br/>test_rg_3.bam</div>#' ) \
+         <( gunzip -c ${expected_gzip} | sed -e 's#"header_filename">.*<br/>test_rg_3.bam#"header_filename"><br/>test_rg_3.bam</div>#' ) \
+    && ( echo "Test PASSED" && exit 0 ) || ( echo "Test FAILED, output file mismatch." && exit 1 )
+    """
+}
+
+
+workflow checker {
+  take:
+    input_file
+    expected_output
+
+  main:
+    fastqc(
+      input_file
+    )
+
+    file_diff(
+      fastqc.out.output,
+      expected_output
+    )
+}
+
+
+workflow {
+  checker(
+    file(params.input_file),
+    file(params.expected_output)
+  )
+}

fastqc/tests/expected/expected.input_file_name.txt

@@ -0,0 +1 @@
+The input file name is README.md

fastqc/tests/input/README.md

@@ -0,0 +1 @@
+This folder contains tiny data files for testing.

fastqc/tests/nextflow.config

@@ -0,0 +1 @@
+includeConfig '../nextflow.config'

fastqc/tests/test-job-1.json

@@ -0,0 +1,7 @@
+{
+    "input_file": "input/test_rg_3.bam",
+    "expected_output": "expected/expected.test_rg_3_fastqc.out.gz",
+    "publish_dir": "outdir",
+    "cpus": 1,
+    "mem": 0.5
+}

fastqc/tests/wfpr_modules

@@ -0,0 +1 @@
+../wfpr_modules

fastqc/wfpr_modules

@@ -0,0 +1 @@
+../wfpr_modules