Skip to content

Whole Genome Sequencing — Germline

Domain key: wgs

Germline WGS detects inherited variants across the entire genome. Codaris builds GATK4 best-practice DSL2 pipelines covering QC → alignment → duplicate marking → BQSR → variant calling.

Step Tool Notes
QC FastQC Per-read quality metrics
Alignment bwa-mem2 2× faster than BWA with identical accuracy
BAM processing SAMtools Sort, index, flagstat
Duplicate marking GATK4 MarkDuplicates Required before variant calling
Base recalibration GATK4 BaseRecalibrator + ApplyBQSR Corrects systematic base quality errors
Variant calling GATK4 HaplotypeCaller Outputs gVCF for joint genotyping

Best Practices Enforced

  • bwa-mem2 is preferred over bwa for speed; results are identical
  • Duplicate marking is non-negotiable — skipping it causes false variant calls
  • BQSR is applied using a known-sites VCF (dbSNP / Mills); Codaris will prompt for the path
  • HaplotypeCaller outputs gVCF (-ERC GVCF) so multi-sample joint genotyping is possible downstream
  • All process labels use process_medium / process_high — memory is never hardcoded

Example Prompts

Build a germline WGS variant calling pipeline for human samples,
aligned to GRCh38, using GATK4 best practices

I have 30x WGS data from 12 patients. Build a pipeline that outputs
per-sample gVCFs ready for joint genotyping with GenomicsDBImport

codaris > Lint my WGS pipeline and check BQSR is applied correctly
          before HaplotypeCaller