A snakemake pipeline to call structural variants from tumor-only ONT data

⚠️ This pipeline is in its early stages. Please use with caution.

Author

Minghao Jiang, jiang01@icloud.com

Tools used

• SV callers

  cuteSV, Sniffles, SVIM, SVision, Severus, NanoSV, NanoVar, Delly, Debreak

• Annotation tools

  AnnotSV, VEP (release/111), SnpEff

• R packages

  vroom, tibble, glue, dplyr, tidyr, purrr, GenomicRanges, stringr, BiocParallel, parallel

• Other tools

  Minimap2, SAMtools, BCFtools, SURVIVOR, vcf2maf (1.6.21), SnpSift, duphold

Pipeline structure

%%{
  init: {
    'theme': 'base',
    'themeVariables': {
      'fontFamily': 'Comic Sans MS',
      'primaryColor': '#F4CE14',
      'primaryTextColor': '#FFFFFF',
      'lineColor': '#EE6983',
      'secondaryColor': '#00B8A9',
      'tertiaryColor': '#FFFFFF'
    }
  }
}%%

---
title: SV calling workflow
---

flowchart TD

  classDef myclass fill:#00B8A9, stroke-width:0px, padding:0px, margin:0px;
  classDef myclass2 fill:#A5DD9B, stroke-dasharray:5 5;

  fastq([FASTQ]) -- Minimap2 --> bam([BAM])
  bam([BAM]) -- cuteSV --> cutesv_vcf([VCF])
  bam([BAM]) -- Sniffles2 --> sniffles2_vcf([VCF])
  cutesv_vcf([VCF]) -- BCFtools --> cutesv_filtered_vcf([filtered VCF])
  sniffles2_vcf([VCF]) -- BCFtools --> sniffles2_filtered_vcf([filtered VCF])
  cutesv_filtered_vcf([filtered VCF]) --- survivor["SURVIVOR"]
  sniffles2_filtered_vcf([filtered VCF]) --- survivor["SURVIVOR"]
  survivor["SURVIVOR"] --> merged_vcf([merged VCF])
  merged_vcf([merged VCF]) -- VEP, AnnotSV, and SnpEff --> annotated_vcf([annotated VCF/TSV])
  annotated_vcf([annotated VCF/TSV]) --> somatic_vcf([somatic SVs])
  annotated_vcf([annotated VCF/TSV]) --> germline_vcf([germline SVs])

  bam([BAM]) -. "other callers (e.g. SVIM)" .-> other_vcfs([VCFs])
  other_vcfs([VCFs]) -. BCFtools .-> other_filtered_vcf([filtered VCFs])

  other_filtered_vcf([filtered VCFs]) -.- survivor["SURVIVOR"]

  survivor:::myclass
  other_vcfs:::myclass2
  other_filtered_vcf:::myclass2

Loading

Getting started

Prerequisites

1. Clone this repo and navigate into it:

   git clone https://github.com/jasonwong-lab/smk_sv.git
   cd smk_sv

   • Follow all steps below after you are in the top dir of this repo.
   • Uncomment all rules in the Snakefile.
   • Check the predefined wildcards_constraints in the Snakefile and modify/delete it if necessary.
   • Using a JSON schema to validate the configuration file might prevent Snakemake from monitoring changes to the parameters. You can comment the validate(config, "config/config.schema.json") in the Snakefile.

2. Install AnnotSV manually.

   • AnnotSV is not included in the image due to its large annotation resources (~ 20GB) that cannot be specified elsewhere.
   • Creating a lock file for each combination of sample and type_sv has been implemented. However, AnnotSV might still encounter errors since it doesn’t support processing multiple files within the same directory. To address this, an additional resource parameter constraint_annotsv=1 has been added to the rule annotate_sv_annotsv to ensure that only one instance of AnnotSV runs at a time. You can modify this parameter in workflow/profile/default/config.yaml where its default is 1.

Configuration

1. Prepare config files:

   1. Copy config/config-test.yaml to config/config.yaml.
      • Adjust the configuration settings according to your project's needs.
      • Specification of important elements:
        • dir_run: working directory where all results will be stored.
        • mapper: dict whose keys are names of mappers and values (boolean) indicate whether perform mapping or not. Only the first mapper will be used. When a mapper is specified and its value is false, no mapping by this mapper will be performed, but its results will be used in the following steps.
        • callers: dict whose keys are names of callers and values (boolean) indicate whether perform SV calling using this caller or not. When a caller is specified and its value is false, no SV calling by this caller will be performed, but its results will be used in the following steps.
        • types_sv: SV types to be called. BND indicates translocations.
        • threads: number of CPUs of each rule to be used.
        • ...
   2. Copy workflow/profiles/default/config-test.yaml to workflow/profiles/default/config.yaml.
      • Bind directories you need in the container.
      • Change the number of CPUs you prefer.
      • Modify/add/delete other parameters of this snakemake pipeline.

2. Prepare sample data:

   1. Copy config/pep/samples-test.csv to config/pep/samples.csv, and update sample_name in the csv.
   2. Copy config/pep/config-test.yaml and config/pep/config.yaml. More information please see Portable Encapsulated Projects (PEP).

3. Set up Conda environments:

   snakemake --conda-create-envs-only

Execution

• Run the pipeline locally:

  snakemake

• Run the pipeline on a cluster: If you want to run this pipeline on a cluster (e.g., SLURM, or PBS), you should customise your own profile and place it into ~/.config/snakemake/, and then run the pipeline with the profile you have set as a parameter:

  snakemake --profile <your_profile_name>

  Or run the pipeline with the profile you have set as an environment variable:

  export SNAKEMAKE_PROFILE=<your_profile_name>
  snakemake

  You can refer to the profile I have been using at workflow/profiles/mycluster, or turn to snakemake websites.

Note for Cluster Users

If you are using a cluster that does not support Singularity well, please switch to the without_docker branch. This branch is tailored for environments where containers might not be the best option.

git checkout without_docker

License

Codes here are licensed under the GNU General Public License v3.