Alignment and Calling

Overall description of VarSome's pipelines

VarSome Clinical uses Sentieon's implementation of bwa-mem for aligning and either gatk or Sentieon's implementation of gatk for variant calling [1]. You can choose which one you want, but we recommend Sentieon's since it's much faster.

  • Sentieon: = Sentieon (bwa-mem) aligner & Sentieon caller
  • GATK: Sentieon (bwa-mem) aligner & gatk caller
  • tumor/somatic samples: Sentieon (bwa-mem) aligner & TNSeq/lofreq caller (see below)

Default Workflow

  1. Adapter trimming is normally done before the FASTQ file is generated. The FASTQ files VarSome gets are therefore already trimmed. Low quality bases are ignored by the variant caller (soft clipping).
  2. Read mapping
  3. Indel realignment
  4. Base-quality recalibration
  5. Variant calling

Joint calling

The key point is that VarSome Clinical doesn't actually have to call variants on all samples together to perform a joint analysis. We have developed a workflow that allows us to decouple the initial identification of potential variant sites (i.e. variant calling) from the genotyping step, which is the only part that really needs to be done jointly. This is the so-called "GVCF workflow", which utilizes a GVCF intermediate to allow scaling joint calling efficiently and conveniently [2].

Tumor and Somatic Samples

  • For matched tumor-normal samples, we use Sentieon's TNSeq tool.
  • For single tumor samples without a matched normal, we use the lofreq variant caller.

MuTect pipeline

Instead of MuTect pipeline, our VarSome Clinical uses by default the lofreq caller for tumor samples. However, MutTect is also available. If you need this pipeline, please get in touch with us.

Related articles