This pipeline implements the GATK best practices variant calling workflow (v3.6) for single samples. It's major output are a recalibrated BAM file, gVCF and VCF files (per sample).
See cfg/references.yaml for references used by default (also refer
to option --references-cfg)
If you are interested in more specialized analyses (e.g. cohort or trio), please reuse the gVCF file (or recalibrated BAM file) produced by this workflow.
Settings are automatically changed according to the given sequencing type, in particular 'base quality recalibration' (BQSR) and 'variant quality recalibration' (VQSR) as explained in the following sections.
BQSR is skipped for targeted resequencing. For an explanation see this GAK article, or this GATK forum post.
Hard variant filtering is always applied, following the recommendations in this this GATK howto and this GATK guide. Note, it is recommended to optimize your filtering settings on a per sample basis (which obviously cannot be part of any automated workflow).
Variant quality recalibration (VQSR) is run only for WGS data. The reason for only running VQSR for WGS data is explained here: Which training sets / arguments should I use for running VQSR? (last updated 2016-08-30):
in order to achieve the best exome results one needs to use an exome SNP and/or indel callset with at least 30 samples.
Note, VQSR can fail for perfectly valid reasons (e.g. too few uniq variants). In such cases a fake vcf file will be produced which contains one line indicating the problem. In such cases, resort to the hard filtered files.
For targeted and whole exome sequencing an interval padding of 100 (or as defined in cfg/params.yaml) will be used as well.
- BAM files:
{sample}.bwamem.dedup.bqsr.bamor{sample}.bwamem.dedup.bam, depending on whether BQSR was run or not (see above) - gVCF:
{bam}.concat.g.vcf.gz - Hard-filtered snp/indel (vartype) calls:
{bam}.gt.{vartype}_hfilter.vcf.gzand with annotation:{bam}.gt.{vartype}_hfilter.snpeff.vcf.gz - Recalibrated snp/indel (vartype) calls (for WGS only):
{bam}.gt.{vartype}_vqsr.vcf.gzand with annotation:{bam}.gt.{vartype}_vqsr.snpeff.vcf.gz
- Call variants with HaplotypeCaller (last updated 2016-02-11)
- Recalibrate variant quality scores = run VQSR (last updated 2014-12-17)
- Apply hard filters to a call set (last updated 2016-07-28)