GATK Variant Discovery Pipeline

Background

As the technology allows regularly producing sequencing data in a high-throughput manner, genome-wide variant discovery becomes routine for many applications, including genotyping of variants for their phenotypic connection. Single nucleotide polymorphisms (SNPs) and small insertion–deletions (indels) are two common forms of genomic variants. The SNP, as the simplest molecular marker, has been widely used to connect genetic variation with phenotypic variation of biological traits, such as human diseases and plant architectures [1,2]. The reference-based SNP calling is the process by which DNA polymorphisms between sequencing reads and the reference are identified based on their alignments [3]. Algorithms and software packages have been developed for SNP calling. Simple SNP calling methods determine read counts at each polymorphic site and set a cutoff of read counts for SNP identification [4]. However, the power of SNP discovery using such methods is subject to sequencing depths. More sophisticated approaches incorporate the uncertainty of SNP calling in a probabilistic framework, such as SOAP2, SAMtools, or Genome Analysis ToolKits (GATK) [5–7]. Among them, GATK is the industry standard for discovering SNPs and small indels using sequencing data from genomic DNA and RNA-seq [7]. Although GATK was originally developed for human sequencing data, it has evolved to handle genome data from other organisms with different levels of ploidy [8]. Here, we present a variant calling pipeline using whole-genome sequencing (WGS) data based on GATK. The pipeline includes reads mapping, variant calling, and variant filtering.