HOME  ›   pipelines
If your question is not answered here, please email us at:  ${email.software}

10x Genomics
Chromium Genome & Exome

Using GATK with Long Ranger

We recommend running longranger with GATK for more accurate calling of SNPs and indels. You must already have a working version of GATK installed on your system. Support is also provided for using a bundled build of Freebayes. Long Ranger uses GATK's HaplotypeCaller mode. See the GATK installation instructions

Performance Comparison

The GATK mode of Long Ranger yields substantially more accurate indel calls. Below are ROC curves reported by hap.py for 30x WGS Chromium data on NA12878, against the Genome In a Bottle v3.2.2 reference variants. GATK improves indel sensitivity by ~5 percentage points, while improving specificty.

gatk vs freebayes snp roc curve gatk vs freebayes indel roc curve

Version Compatibility

Long Ranger VersionGATK Version Supported
2.2 and later3.3-4.0, excluding 3.6
2.1 and earlier3.3-3.8, excluding 3.6

Configuring GATK

To use GATK, please note the following:

  1. You must already have a working version of GATK installed on your system; the Long Ranger software does not include the GATK distribution.
  2. You must use a version of GATK that is compatible with your version of Long Ranger; see the Version Compatibility table, above. Note that GATK version 3.6 has a crashing bug in the 'single-haplotype' mode, which Long Ranger uses, and so is not compatible..
  3. If you are using a custom reference, longranger mkref will instruct you on how to create a genome.dict file. You will need this file to use GATK.

You can instruct longranger to use GATK as its variant caller by passing the --vcmode=gatk:/path/to/GenomeAnalysisTK.jar parameter. For example, if your GenomeAnalysisTK.jar jarfile is located in /usr/local/gatk/GenomeAnalysisTK.jar,

$ longranger wgs --id=sample345 \
                 --sex=female \
                 --fastqs=/home/jdoe/runs/HAWT7ADXX/outs/fastq_path \
                 --indices=SI-GA-A1 \
                 --reference=/opt/refdata-hg19-2.1.0 \
                 --targets=/home/jdoe/runs/agilent_exome.bed \
                 --vcmode=gatk:/usr/local/gatk/GenomeAnalysisTK.jar \

Aside from the stage where SNPs and indels are called, Long Ranger will behave the same whether using GATK or Freebayes.