How to use GATK best practices for calling variants?

Set Up Your Environment

• Ensure you have the latest version of GATK installed. You can download it from the GATK website.

 

• Install Java Runtime Environment (recommended version 8 or later).

 

• Have access to necessary reference files like the human genome reference sequence (e.g., hg19 or hg38), and known sites of variation (e.g. dbSNP, Mills and 1000G gold standard indels).

Prepare Your Data

• Obtain high-quality FASTQ files from your sequencer, ensuring the data is correctly paired (for paired-end reads).

 

• Use BWA (Burrows-Wheeler Aligner) to map the reads to the reference genome. The command line might look like: bwa mem -t 4 reference.fa sample_R1.fastq sample_R2.fastq > alignment.sam.

 

• Convert the SAM files to BAM, and sort them using samtools: samtools sort alignment.sam -o sorted\_alignment.bam.

Mark Duplicates

• Use GATK's MarkDuplicates to mark duplicate reads in the BAM file: gatk MarkDuplicates -I sorted\_alignment.bam -O deduped.bam -M metrics.txt.

 

• This step helps to reduce bias in variant calling by eliminating PCR duplicates.

Base Quality Score Recalibration (BQSR)

• Create a recalibration table using known sites of variation: gatk BaseRecalibrator -I deduped.bam -R reference.fa --known-sites dbsnp.vcf --known-sites mills.vcf -O recal\_data.table.

 

• Apply the recalibration to your BAM file: gatk ApplyBQSR -R reference.fa -I deduped.bam --bqsr-recal-file recal\_data.table -O recalibrated.bam.

Variant Calling

• Use HaplotypeCaller to call variants on the recalibrated data: gatk HaplotypeCaller -R reference.fa -I recalibrated.bam -O raw\_variants.vcf.

 

• This generates a VCF file containing raw variant calls.

Variant Filtering

• Use VariantFiltration to apply filters to your raw VCF file, based on quality metrics: gatk VariantFiltration -R reference.fa -V raw_variants.vcf --filter-expression "QUAL < 30.0" --filter-name "LowQual" -O filtered_variants.vcf.

 

• To refine results for SNPs, use GATK’s VariantRecalibrator for SNPs and indels if additional annotations are available.

Evaluate and Validate Variants

• Ensure the integrity of your variant calls by comparing against a gold-standard dataset if available.

 

• Use tools such as vcftools or bcftools to analyze your VCF file and extract insights or generate reports.

Document and Archive

• Keep detailed records of the parameters and versions of software used in your variant calling process for replication and review purposes.

 

• Store your results, along with reference data and BAM files, in a secure location in compliance with data management policies.