Stian Soiland-Reyes

eScience lab, The University of Manchester

@soilandreyes

https://orcid.org/0000-0001-9842-9718

https://slides.com/soilandreyes/

Cancer Research UK, Manchester Institute
Alderly Edge, 2019-09-18

This work has been done as part of the BioExcel CoE (www.bioexcel.eu), a project funded by the European Union contracts H2020-INFRAEDI-02-2018-823830, H2020-EINFRA-2015-1-675728

CWL Sharing reproducible computational analyses

Findable

Accessible

Interoperable

Reusable

F1. (meta)data are assigned a globally unique and persistent identifier

 

F2. data are described with rich metadata (defined by R1 below)

 

F3. metadata clearly and explicitly include the identifier of the data it describes

 

F4. (meta)data are registered or indexed in a searchable resource

 

To be Findable:

A1. (meta)data are retrievable by their identifier using a
    standardized communications protocol
 

  A1.1 the protocol is open, free, and universally implementable


  A1.2 the protocol allows for an authentication and authorization
           procedure, where necessary

 

A2. metadata are accessible, even when the data are no longer available

To be Accessible:

I1. (meta)data use a formal, accessible, shared, and
    broadly applicable language for knowledge representation.

 

I2. (meta)data use vocabularies that follow FAIR principles

 

I3. (meta)data include qualified references to other (meta)data

To be Interoperable:

R1. meta(data) are richly described with a plurality of
    accurate and relevant attributes

 

  R1.1. (meta)data are released with a clear and accessible data usage license
 

  R1.2. (meta)data are associated with detailed provenance

 

  R1.3. (meta)data meet domain-relevant community standards

To be Reusable:

Inner FAIR

Workflows

Automation

– Automate computational aspects
– Repetitive pipelines, sweep campaigns

Scalingcompute cycles

– Make use of computational infrastructure
– Handle large data

Abstraction⁠—people cycles

– Shield complexity and incompatibilities
– Report, re-usue, evolve, share, compare
– Repeat—Tweak—Repeat
– First-class commodities

Provenancereporting

– Capture, report and utilize log and data lineage
– Auto-documentation
– Tracable evolution, audit, transparency
– Reproducible science

Findable

Accessible

Interoperable

Reusable

(Reproducible)

Why use workflows?

cwlVersion: v1.0
class: Workflow
inputs:
  inp: File
  ex: string

outputs:
  classout:
    type: File
    outputSource: compile/classfile

steps:
  untar:
    run: tar-param.cwl
    in:
      tarfile: inp
      extractfile: ex
    out: [example_out]

  compile:
    run: arguments.cwl
    in:
      src: untar/example_out
    out: [classfile]

Nature 573, 149-150 (2019)
https://doi.org/10.1038/d41586-019-02619-z

Building CWL workflows

cwlVersion: v1.0
class: Workflow

inputs:
  toConvert: File


outputs:
  converted: 
    type: File
    outputSource: convertMethylation/converted
  combined: 
    type: File
    outputSource: mergeSymmetric/combined


steps:
  convertMethylation:
    run: interconverter.cwl
    in:
      toConvert: toConvert
    out: [converted]
  mergeSymmetric:
    run: symmetriccpgs.cwl
    in:
      toCombine: convertMethylation/converted
    out: [combined]
cwlVersion: v1.0
class: CommandLineTool
baseCommand: interconverter.sh
hints:
  - class: DockerRequirement
    dockerPull: "quay.io/neksa/screw-tool"
arguments: ["-d", $(runtime.outdir)]
inputs:
  toConvert:
    type: File
    inputBinding:
      prefix: -i
outputs:
  converted:
    type: File
    outputBinding:
      glob: "*.meth"
cwlVersion: v1.0
class: CommandLineTool
baseCommand: symmetriccpgs.sh
arguments: ["-d", $(runtime.outdir)]
hints:
  - class: DockerRequirement
    dockerPull: "quay.io/neksa/screw-tool"

inputs:
  toCombine:
    type: File
    inputBinding:
      prefix: -i
outputs:
  combined:
    type: File
    outputBinding:
      glob: "*.sym"
cwlVersion: v1.0
class: Workflow
label: EMG QC workflow, (paired end version). Benchmarking with MG-RAST expt.

requirements:
 - class: SubworkflowFeatureRequirement
 - class: SchemaDefRequirement
   types: 
    - $import: ../tools/FragGeneScan-model.yaml
    - $import: ../tools/trimmomatic-sliding_window.yaml
    - $import: ../tools/trimmomatic-end_mode.yaml
    - $import: ../tools/trimmomatic-phred.yaml

inputs:
  reads:
    type: File
    format: edam:format_1930  # FASTQ

outputs:
  processed_sequences:
    type: File
    outputSource: clean_fasta_headers/sequences_with_cleaned_headers

steps:
  trim_quality_control:
    doc: |
      Low quality trimming (low quality ends and sequences with < quality scores
      less than 15 over a 4 nucleotide wide window are removed)
    run: ../tools/trimmomatic.cwl
    in:
      reads1: reads
      phred: { default: '33' }
      leading: { default: 3 }
      trailing: { default: 3 }
      end_mode: { default: SE }
      minlen: { default: 100 }
      slidingwindow:
        default:
          windowSize: 4
          requiredQuality: 15
    out: [reads1_trimmed]

  convert_trimmed-reads_to_fasta:
    run: ../tools/fastq_to_fasta.cwl
    in:
      fastq: trim_quality_control/reads1_trimmed
    out: [ fasta ]

  clean_fasta_headers:
    run: ../tools/clean_fasta_headers.cwl
    in:
      sequences: convert_trimmed-reads_to_fasta/fasta
    out: [ sequences_with_cleaned_headers ]


$namespaces:
 edam: http://edamontology.org/
 s: http://schema.org/
$schemas:
 - http://edamontology.org/EDAM_1.16.owl
 - https://schema.org/docs/schema_org_rdfa.html

s:license: "https://www.apache.org/licenses/LICENSE-2.0"
s:copyrightHolder: "EMBL - European Bioinformatics Institute"

Finding tools

Containers and packages

FROM ubuntu:14.04
MAINTAINER Stian Soiland-Reyes <soiland-reyes@cs.manchester.ac.uk>

# Install apache/PHP for REST API
RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y \
	apache2 php5 memcached php5-memcached php5-memcache php5-curl \
        supervisor \
        unzip wget 
RUN a2enmod rewrite

RUN rm -rf /var/www/html
#Install Linked Data API
RUN wget https://github.com/openphacts/OPS_LinkedDataApi/archive/1.5.0.zip -O /tmp/api.zip && \
    cd /tmp && \
    unzip api.zip && \
    mv /tmp/OPS* /var/www/html

RUN sed -i '/<\/VirtualHost/ i\ <Directory /var/www/html/>\n  AllowOverride All\n </Directory>' /etc/apache2/sites-enabled/000-default.conf
#RUN cat /etc/apache2/apache2.conf | tr "\n" "|||" | \
#      sed 's,\(<Directory /var/www/html/>[^<]*\)AllowOverride None\([^<]*</Directory>\),\1AllowOverride All\2,' | \
#      sed 's/|||/n/g' >/tmp/apache2 && \
#    mv /tmp/apache2 /etc/apache2/apache2.conf
RUN mkdir /var/www/html/logs /var/www/html/cache && \
    chmod 777 /var/www/html/logs /var/www/html/cache && \
    chown -R www-data:www-data /var/www/html

Over 18,000 CWL descriptions on GitHub
(of which ~5,500 workflows)

 

https://view.commonwl.org/workflows

CWL Features

Interoperability

.. and portability

cwltool: local (Linux, OS X, Windows) — reference implementation

Arvados: AWS, GCP, Azure, Slurm, SaaS

Toil: local, AWS, Azure, GCP, SGE, LSF, Mesos, OpenStack, Slurm, PBS/Torque

CWL-Airflow: Linux, OS X

REANA: Kubernetes, CERN OpenStack
cromwell: local, HPC, Google, HtCondor

CWLEXEC: IBM Spectrum LSF

Which CWL engine runs where?

Galaxy: local (Linux), PBS/Torque, PBS Pro, LSF, Grid Engine, HTCondor, Slurm

Rabix Bunny: Local(Linux, OS X), GA4GH TES, SaaS

cwl-tes: Local, AliCloud, GCP, AWS, HTCondor, Grid Engine, PBS/Torque, Slurm

XENON: local, ssh, SLURM, Torque, Grid Engine

Consonance: AWS, OpenStack, Azure

Calrissian: Kubernetes

(..)

Portable command line tools

#!/usr/bin/env cwl-runner

cwlVersion: v1.0

class: CommandLineTool

baseCommand: ["gmx", "pdb2gmx"]

arguments: ["-o", "processed.gro"]

inputs:
  pdb:
    type: string
    inputBinding:
      prefix: -f
  water:
    type: File
    inputBinding:
      prefix: -water
    default: spce

outputs:
  processed_gro:
    type: File
    outputBinding:
      glob: processed.gro

Portable command line tools

Where to find command line tools?

cwlVersion: v1.0
class: CommandLineTool
baseCommand: node
hints:
  DockerRequirement:
    dockerPull: gromacs/gromacs:2018.4

class: CommandLineTool
hints:
  SoftwareRequirement:
    packages:
      samtools:
        version: [ "0.1.19" ]

baseCommand: ["samtools", "index"]
#..

Finding the tool

module load samtools/0.1.19
apt-get install samtools=0.1.19*
conda install samtools=0.1.19

Let's add some identifiers!

hints:
  SoftwareRequirement:
    packages:
    - package: gromacs
      version:
      - '2018.4'
      specs:
      - https://packages.debian.org/gromacs
      - https://anaconda.org/bioconda/gromacs
      - https://bio.tools/gromacs
      - https://identifiers.org/rrid/RRID:SCR_014565
      - https://hpc.example.edu/modules/gromacs/2018.4
  DockerRequirement:
    dockerPull: gromacs/gromacs:2018.4

Going beyond CWL

#!/usr/bin/env cwl-runner

cwlVersion: v1.0
class: CommandLineTool
baseCommand: echo

requirements:
  InlineJavascriptRequirement: {}

inputs: []
outputs:
  example_out:
    type: stdout
stdout: output.txt
arguments:
  - prefix: -A
    valueFrom: $(1+1)
  - prefix: -B
    valueFrom: $("/foo/bar/baz".split('/').slice(-1)[0])
  - prefix: -C
    valueFrom: |
      ${
        var r = [];
        for (var i = 10; i >= 1; i--) {
          r.push(i);
        }
        return r;
      }

JavaScript expressions

#!/usr/bin/env cwl-runner
cwlVersion: v1.0
class: CommandLineTool

requirements:
  DockerRequirement:
    dockerPull: lukasheinrich/dummyanalysis
  ShellCommandRequirement: {}

baseCommand: /bin/bash

arguments:
  - valueFrom: |
      source /usr/local/bin/thisroot.sh
      cat $(runtime.outdir)/input_list | xargs hadd merged.root
    prefix: -c
   
# (..)

ShellCommandRequirement

#!/usr/bin/env cwl-runner
#
# Example score submission file
#
cwlVersion: v1.0
class: CommandLineTool

requirements:
  - class: InlineJavascriptRequirement
  - class: InitialWorkDirRequirement
    listing:
      - entryname: score.py
        entry: |
          #!/usr/bin/env python
          import synapseclient
          import argparse
          import os
          import json
          parser = argparse.ArgumentParser()
          parser.add_argument("-f", "--submissionfile", required=True, help="Submission File")
          # ..
          args = parser.parse_args()
          score = 3
          prediction_file_status = "SCORED"
          result = {'score':score, 'prediction_file_status':prediction_file_status}
          with open(args.results, 'w') as o:
            o.write(json.dumps(result))

baseCommand: python

arguments:
  - valueFrom: score.py
  - valueFrom: $(inputs.inputfile.path)
    prefix: -f


inputs:
  - id: inputfile
    type: File


outputs:
  - id: results
    type: File
    outputBinding:
      glob: results.json

Example: Embedding Python script

Encouraging best practice

Not just files

Secondary files (e.g. *.bam.bai)

Inline files/strings

Directories

Arrays of files

Enums

JSON records (string, int, array, dict)

 

CWL implementations can decide how to handle data management, e.g. data store or file copying

Dataflow features

Optional inputs (tool reuse)

Sub-workflows (reuse workflow as a tool)

Task Parallelization (step execute when data is ready)

Scattering (multiple tasks from single array)

Merge (multiple inputs to single array)

Provenance

Who ran it?

When did it run?

Where did it run?

What workflow ran?

Which tool versions?

What data was created?

 

Copyright © 2013 W3C® (MIT, ERCIM, Keio, Beihang), All Rights Reserved. W3C liability, trademark and document use rules apply.

PROV Model Primer

W3C Working Group Note 30 April 2013

Khan et al,
Sharing interoperable workflow provenance: A review of best practices and their practical application in CWLProv
GigaScience (in print)
https://doi.org/10.5281/zenodo.3336124


(venv3) stain@biggie:~/src/cwlprov-py/test/nested-cwlprov-0.3.0$ cwlprov run
2018-08-08 22:44:06.573330 Flow 39408a40-c1c8-4852-9747-87249425be1e [ Run of workflow/packed.cwl#main 
2018-08-08 22:44:06.691722 Step 4f082fb6-3e4d-4a21-82e3-c685ce3deb58   Run of workflow/packed.cwl#main/create-tar  (0:00:00.010133)
2018-08-08 22:44:06.702976 Step 0cceeaf6-4109-4f08-940b-f06ac959944a * Run of workflow/packed.cwl#main/compile  (unknown duration)
2018-08-08 22:44:12.680097 Flow 39408a40-c1c8-4852-9747-87249425be1e ] Run of workflow/packed.cwl#main  (0:00:06.106767)
Legend:
[ Workflow start
* Nested provenance, use UUID to explore: cwlprov run 0cceeaf6-4109-4f08-940b-f06ac959944a
] Workflow end

(venv3) stain@biggie:~/src/cwlprov-py/test/nested-cwlprov-0.3.0$ cwlprov run 0cceeaf6-4109-4f08-940b-f06ac959944a
2018-08-08 22:44:06.607210 Flow 0cceeaf6-4109-4f08-940b-f06ac959944a [ Run of workflow/packed.cwl#main 
2018-08-08 22:44:06.707070 Step 83752ab4-8227-4d4a-8baa-78376df34aed   Run of workflow/packed.cwl#main/untar  (0:00:00.008149)
2018-08-08 22:44:06.718554 Step f56d8478-a190-4251-84d9-7f69fe0f6f8b   Run of workflow/packed.cwl#main/argument  (0:00:00.532052)
2018-08-08 22:44:07.251588 Flow 0cceeaf6-4109-4f08-940b-f06ac959944a ] Run of workflow/packed.cwl#main  (0:00:00.644378)
Legend:
[ Workflow start
] Workflow end
stain@biggie:~/src/cwlprov-py/test/nested-cwlprov-0.3.0$ cwlprov outputs 4f082fb6-3e4d-4a21-82e3-c685ce3deb58 --format=files
Output tar:
data/c0/c0fd5812fe6d8d91fef7f4f1ba3a462500fce0c5

stain@biggie:~/src/cwlprov-py/test/nested-cwlprov-0.3.0$ tar tfv `cwlprov -q outputs 4f082fb6-3e4d-4a21-82e3-c685ce3deb58 --format=files`
-rw-r--r-- stain/stain     115 2018-08-08 23:44 Hello.java

Inspecting step runs

$ cwlprov --help
usage: cwlprov [-h] [--version] [--directory DIRECTORY] [--relative]
            [--absolute] [--output OUTPUT] [--verbose] [--quiet] [--hints]
            [--no-hints]
            {validate,info,who,prov,inputs,outputs,run,runs,rerun,derived,runtimes}
            ...

cwlprov explores Research Objects containing provenance of Common Workflow
Language executions. <https://w3id.org/cwl/prov/>

commands:
{validate,info,who,prov,inputs,outputs,run,runs,rerun,derived,runtimes}
    validate            Validate the CWLProv Research Object
    info                show research object Metadata
    who                 show Who ran the workflow
    prov                export workflow execution Provenance in PROV format
    inputs              list workflow/step Input files/values
    outputs             list workflow/step Output files/values
    run                 show workflow Execution log
    runs                List all workflow executions in RO
    rerun               Rerun a workflow or step
    derived             list what was Derived from a data item, based on
                        activity usage/generation
    runtimes            calculate average step execution Runtimes

Work in progress

 

towards CWL 2.x

Conditional branching (~switch statement)

Looping

 

usability

CWL integrations: Galaxy, KNIME

Ansible scripts for installation

Text editor language support (benten)

 

flexibility

alternative script-like grammar (cwl-ex)

findability

Federated workflow repository (EOSC-Life)

Tool collection (common-workflow-library)

 

training

Virtual Training (BioExcel)

Introductory user guide

Join the CWL community

2019-09-18 CWL - Sharing reproducible computational analyses

By Stian Soiland-Reyes

2019-09-18 CWL - Sharing reproducible computational analyses

Seminar at Cancer Research UK, Manchester 2019-09-18

  • 176
Loading comments...

More from Stian Soiland-Reyes