Skip to content

Sarek

Sarek is a pipeline built using Nextflow designed to detect variants in whole or targeted sequencing data. It works on any species with a reference genome and allows for germline, tumor only and tumor-normal pair variant calling

Description

Sarek workflow consists of the following steps:

  • Preprocessing (based on GATK4 Best Practices): includes sequencing quality control, reads alignment (BWA), mark duplicates and base recalibration.
  • Variant calling. You can choose any caller, but here are some recommendations:
    • HaplotypeCaller for germline SNVs and indels.
    • Mutect2 and Strelka for somatic SNVs and indels.
    • Manta for structural variants (SVs).
    • ASCAT for copy number variants (CNVs)
  • Annotation: you can choose snpEff, VEP or merge both.

Note sarek allows to run the pipeline starting from any step.

Installation

You can follow the instructions in Sarek's webpage. Take this into account if you plan to use sarek in the cluster:

  1. Create a conda environment to install the latest java version (openjdk specifically).
  2. Preferably, use Singularity as container. It is not necessary to install it as it is already installed in the cluster.

  3. Better download and run the pipeline test specifying the version, entering the -r argument:

    nextflow run nf-core/sarek -r 3.1.2 -profile test,singularity

  4. Create a config file (you can name it sarek.conf) to specify the executor and Singularity cache directory. In this file you can also add any changes in the configuration specification of the pipeline steps (e.g.: memory limit). Example:

    executor {
    name = 'slurm'
    queueSize = 25
}

singularity {
    cacheDir = '/home/$USER/singcache'
}

process {
    withName: 'NFCORE_SAREK:SAREK:BAM_MARKDUPLICATES:GATK4_MARKDUPLICATES' {
        cpus = 28
        memory = 200.GB
    }
}

Usage

To run the pipeline you can follow the instruction in sarek's webpage, first activating the conda environment with java installed. Here is an example of a command to run the analysis from alignment generation to variant annotation, calling germline and somatic SNVs, SVs and CNVs:

conda activate java
nextflow run nf-core/sarek -r 3.1.2 -profile singularity -c /path/to/sarek.conf \
    --input input.csv --genome GATK.GRCh38 --igenomes_base /path/to/igenomes/ \
    --tools 'haplotypecaller,strelka,ascat,manta,mutect2,vep'

To start the pipeline from a different step you will have to include the --step flag in the command. If you only want to generate the alignments, omit the --tools flag. To avoid pipeline freezing, it is highly recommended to use a local version of iGenomes instead of pulling it from AWS during pipeline execution.

Create input.csv

Sarek uses a comma-separated samplesheet with information about the samples to run the pipeline. Here you specify information as the patient-id, sex, status (normal/tumor), etc. In sarek's webpage you can find examples of the mandatory fields required in the samplesheet depending on the pipeline step from which you start the analysis.

Main outputs

If you run the entire pipeline, you will end with many files. From those, we will highlight:

  • In results/preprocessing/recalibrated/ the final alignments in CRAM format per sample.
  • In results/variant_calling/ the VCFs per caller and sample.
  • In results/annotation/ the annotated VCFs per caller and sample.
  • Sarek official webpage
  • As an output example, the data in /workspace/datasets/all_aecc_pediatric was generated using sarek version 3.1.1. Feel free to give it a look if needed.

Reference

  • Raquel Blanco
  • Monica Sanchez
  • Miguel Grau