NIAGADS
  • VCPA
  • Introduction
  • Step 1: Set up the Amazon Web Services (AWS) environment
    • 1.1 Create AWS account
    • 1.2 Configure your computing environment and login to AWS
    • 1.3 Setup a S3 bucket (simple storage solution for AWS) for hosting sequencing data
    • 1.4 Install AWS command line software for accessing S3 bucket via command line interface
    • 1.5 Install StarCluster for AWS instance provisioning (optional)
  • Step 2: Create your tracking database instance
    • Option 1: Setup sample tracking database using Public AMI (recommended)
    • Option 2: Setup sample tracking database using Docker
  • Step 3: Configure your project information in the tracking database
    • 3.1 List all projects in the tracking database
    • 3.2 Create the project in the tracking database
  • Step 4: Upload sequencing data to your S3 bucket
  • Step 5: Configure your samples information in the tracking database
    • 5.1 Input the sample information to the tracking database
    • 5.2 Populate the tracking database with the S3 paths for the samples to be processed
    • 5.3 Populate the tracking database with the designated result folder for each sample to be processed
    • 5.4 Input PCR protocol information into the tracking database
    • 5.5 Add the capture kit information (WES sample only) into the tracking database
    • 5.6 Generate an ID to represent the capture kit information (WES sample only)
  • Step 6: Submit a job to process one whole genome (WGS) / whole exome (WES) sample
    • 6.1 Update vcpa-pipeline bitbucket contents
    • 6.2 Choose which workflow to use
    • 6.3 Enter your AWS credentials into the workflow script
    • 6.4 Launch Amazon EC2 Spot Instances via starcluster
  • Step 7: Review quality metrics of processed data
  • Step 8: Generating Project-level VCF via joint genotyping
  • Optional: Change software versions and dependencies of the VCPA workflow
Powered by GitBook
On this page

Step 8: Generating Project-level VCF via joint genotyping

PreviousStep 7: Review quality metrics of processed dataNextOptional: Change software versions and dependencies of the VCPA workflow

Last updated 6 years ago

After generating the gVCFs, users can follow the steps provided below for joint genotype call for sample level gVCF into a project level VCF. VCPA implements these steps by referencing to the best practices of GATK. Refer to stage 3 of the VCPA pipeline for details.

To summarize: We used TileDB from Intel to combine all the gVCFs then run the GenotypeGVCF from GATK to do the joint genotype calling. After that, we recalibrate the variant quality scores (VQSR) and apply the desired level of recalibration to the SNPs and INDELs in the call set, also from GATK. Last, we separate biallelic and multiallelic variants by bcftools.

1) Combine multiple gVCFs ( from intel/MIT) - combine all the sample level gVCFs using the combinegVCFs function.

2) GenotypedGVCF (GATK3.7) - joint genotype calling on the output from previous step.

3) Variant Quality Score Recalibration (VQSR) on both SNPs and indels (GATK3.7) - this is to recalibrate the SNPs and indels via VQSR scores

4) ApplyRecalibration (GATK3.7) - this is for applying the desired level of recalibration to the SNPs and indels in the call set

5) Separate the Biallelic and Multiallelic variants (bcftools) - QC will handle biallelic and multiallelic variants separately.

https://bitbucket.org/NIAGADS/vcpa-pipeline/src/master/VCPA/stage3/
TileDB