SAREK Workflow Overview

This repository provides scripts, sample tables, and configuration JSONs to run the SAREK pipeline for calling variant from AIL data.

Below is a high-level overview of the workflow.

---

Step 1. Initial Variant Calling

Run SAREK with Strelka, DeepVariant, and FreeBayes:

• Invoking script:
  sarek_invoking_scripts/AIL_sarek.sh

• Config file:
  sarek_configs/ail_params_no_GATK.json

• Input files:
  Fastq files (see sample tables)

This step generates variant calls from three independent callers.

---

Step 2. Generate Consensus "Known Variants"

Create a consensus set of high-confidence variants by intersecting calls from all three variant callers.

• Scripts:
  • get_known_sites/module_make_known_sites_VCF.sh
  • get_known_sites/module_isec_merge_known_sites_VCF.sh

This produces the known_sites VCF used for downstream base quality recalibration and haplotype calling.

---

Step 3. Haplotype Calling with GATK

Run SAREK with GATK HaplotypeCaller, using the consensus known_sites VCF generated in Step 2.

• Invoking script:
  sarek_invoking_scripts/AIL_haplotypcaller.sh

• Config file:
  sarek_configs/ail_params_with_haplotypecaller.json

• Input files:
  BAM files generated in Step 1

This step improves variant calling accuracy by leveraging the known sites.

---

Notes on Parental Analysis

• Parental samples were analyzed without base recalibration (by design).
• Scripts for extracting parental reads are located in:
  sarek/bb_scripts
• ⚠️ These scripts currently require updates and will not run as-is.

---

Workflow Summary

1. Run SAREK with Strelka, DeepVariant, FreeBayes → produce initial variant calls.
2. Intersect variants → generate consensus known_sites VCF.
3. Run SAREK with GATK HaplotypeCaller using consensus known_sites.