ICGC-ARGO Open-access Somatic Variant Filtering Workflow

Introduction

This repository maintains the source code of the ICGC ARGO Open-access Somatic Variant Filtering Workflow. It is a bioinformatics workflow that can be used to filter SNV/InDel VCFs based on predefined genomic regions.

For ICGC-ARGO, these genomic regions are of our most research interests and the variants among these regions can potentially go to be open-access with minimum risk of germline leakage. Please refer to ICGC-ARGO Open-access Regions for details of how we define and generate the regions.

The workflow is built using Nextflow DSLv2, with modules imported from other ICGC ARGO Workflows GitHub repositories. It uses Docker containers making installation trivial and results highly reproducible. Specifically, here are repositories maintaining various tools/modules:

• https://github.com/icgc-argo/data-processing-utility-tools
• https://github.com/icgc-argo/variant-calling-tools/variant-filter
• https://github.com/icgc-argo/nextflow-data-processing-utility-tools

Each Nextflow module (including associated container image which is registered in Quay.io) is strictly version controlled and released independently. To ensure reproducibility the pipeline declares explicitly which specific version of a module is to be imported.

Workflow steps and summary

• download: Download input variant calling metadata and VCF from SONG/SCORE
• metadata parse: Parse the metadata to get original variant calling tool which was used to generate the input VCF
• filter: Perform SNV/InDel variants filtering with Bcftools:view using different filters according to various variant calling tools
• payload generation: Generate SONG metadata for filtered SNV/InDel calls and upload them to SONG/SCORE

Usage

Prerequisite

• Install Nextflow (>=20.10).
• Install Docker for full pipeline reproducibility.

Reference

You will need to provide a BED file containing genomic regions about the VCFs you would like to filter before running the workflow. Use this parameter to specify its location.

--regions_file '[path to genomic region file]'

Running the workflow

The typical command for running the workflow of verion 0.4.0 is as follows:

nextflow run icgc-argo-workflows/open-access-variant-filtering -r 0.4.0 -params-file <your_params_file.json>

This will launch the workflow with the docker configuration profile. See below for more information about profiles.

Note that the pipeline will create the following files in your working directory:

work            # Directory containing the nextflow working files
results         # Output results (configurable by `publish_dir`, see below)
.nextflow_log   # Log file from Nextflow
# Other nextflow hidden files, eg. history of pipeline runs and old logs.

The workflow could be setting to run in two modes: local mode and RDPC mode.

Local mode

For local mode, both the SNV/InDel variant calling VCFs/Index and related metadata need to be in place at local. You will also need to spcify the path to keep the outputs.

• Inputs

--analysis_metadata '[path to metadata file]'
--vcf_file '[path to SNV/InDel variant calling VCF]'
--regions_file '[path to local genomic region file]' 
--publish_dir '[path to keep the outputs]'

• Outputs

You will find the workflow outputs under the specified publish_dir directory. For instance:

OpenFilterWf_pGenVar
    ├── 51a7ee15-f114-468c-80f0-a1f5bd88e100.variant_processing.payload.json
    └── out
        ├── TEST-PR.DO250183.SA610229.wgs.20200513.sanger-wgs.somatic.snv.open-filter.vcf.gz
        └── TEST-PR.DO250183.SA610229.wgs.20200513.sanger-wgs.somatic.snv.open-filter.vcf.gz.tbi

RDPC mode

For RDPC mode, SONG/SCORE services need to be available and you have appropriate API token. You also need to make sure the referece files have been pre-staged into RDPC NFS.

• Inputs

--study_id '[ICGC-ARGO study ID]'
--analysis_id '[ICGC-ARGO SONG variant_calling analysis ID]'
--song_url": '[ICGC-ARGO SONG server URL]'
--score_url": '[ICGC-ARGO SCORE server URL]'
--regions_file '[path to ICGC-ARGO RDPC NFS genomic region file]' 

• Outputs

The generated outputs are uploaded to SONG/SCORE automatically. You will need use the output analysis_id to retrieve both the metadata and filtered VCFs.

Parameters

Argument name	Description	Default value	Requirement
study_id	ICGC-ARGO study ID	null	Required in RDPC mode
analysis_id	ICGC-ARGO SONG analysis ID	null	Required in RDPC mode
analysis_metadata	Local SONG analysis metadata	null	Required in local mode
vcf_file	Local SNV/InDel variant calling VCF. The index file is expected to be under the same folder	null	Required in local mode
region_file	BED file to define ARGO open access regions	null	Required
api_token	SONG/SCORE API token	null	Required in RDPC mode
song_url	SONG server URL	null	Required in RDPC mode
score_url	SCORE server URL	null	Required in RDPC mode
output_type	Output - compressed BCF (b) - uncompressed BCF (u) - compressed VCF (z) - uncompressed VCF (v)	z	Optional
apply_filters	Skip sites where FILTER column does not contain any of the strings listed in LIST.	{ 'CaVEMan': "PASS", 'Pindel': "PASS", 'GATK:Mutect2': "PASS" }	Optional
include	Select sites for which the expression is true	{ 'CaVEMan': "INFO/CLPM=0 && INFO/ASRD>=0.93", 'Pindel': "", 'GATK:Mutect2': "" }	Optional
exclude	Exclude sites for which the expression is true	{ 'CaVEMan': "", 'Pindel': "", 'GATK:Mutect2': "" }	Optional
open	If true(default), then the output files are set with open access	true	Optional
cleanup	If true(default), then clean up the inputs/temporary/outputs under the work_dir. Skip cleanup when running in local mode.	true	Optional
cpus	Set requirements for number of CPUs for all processes	1	Optional
mem	Set requirements for memory in GB for all processes	1	Optional

Custom configuration

Resource requests

Although the default requirements set within the workflow will hopefully work for most people and with most input data, you may find that you want to customise the compute resources that the workflow requests. Each step in the workflow has a default set of requirements for number of CPUs and memory. You can easily change the default values by setting them in the params-file.

For instance, if you want to change the default settings for download step, you can specify parameters for download step in the params-file as follow:

"download": 
  {
    "song_cpus": 2,
    "song_mem": 2,
    "score_cpus": 3,
    "score_mem": 8
  }

Nextflow will overwrite the default values of settings that you provide via the provided params-file.

Tool specific options

For the ultimate flexibility, we are using Nextflow DSL2 modules in a way where it is possible for both developers and users to change tool-specific command-line arguments (e.g. using a different filtering params for filter step) as well as publish_dir options (e.g. saving files generated by the filter step to specific folder). In most cases, as a user you won't have to change the default options set by the workflow developer(s), however, there may be some cases where providing a custom params file can improve the behaviour or increase the reusability of the workflow.

For instance, if you want to change the default filtering settings for filter step, you can overwrite the defaults by setting param filter in the params-file as follow:

"filter":
  {
    'cpus': 2,
    'mem': 3,
    'publish_dir': 'my_dir',
    'regions_file': 'my_regions_file',
    'apply_filters': {
      'CaVEMan': "VUM", 
      'Pindel': "FF004", 
      'GATK:Mutect2': "strand_bias"
    },
    'include': {
      'CaVEMan': "INFO/CLPM>0 && INFO/ASRD>=0.83", 
      'Pindel': "", 
      'GATK:Mutect2': ""
    },
    'exclude': {
      'CaVEMan': "", 
      'Pindel': "", 
      'GATK:Mutect2': ""
    },
    'output_type': 'b'
  }