Overview
Strelka2 is a small variant caller designed for analyzing germline and somatic variations in DNA sequencing data. It uses a tiered haplotype model for improved accuracy in germline variant calling, adaptively selecting between assembly and alignment-based haplotyping. For somatic variant calling, Strelka2 accounts for potential tumor cell contamination in normal samples. The tool incorporates a mixture-model indel error estimation method to enhance robustness against indel noise. A final variant re-scoring step, utilizing random forest models trained on call quality features, is applied to improve precision. Strelka2 accepts BAM or CRAM files as input and reports variants in VCF 4.1 format, supporting gVCF conventions for germline variant reporting. Integration with Manta structural variant caller is recommended for optimal somatic indel performance.