Somatic small-variant calling methods in Illumina DRAGEN™ Secondary Analysis

Abstract We present the DRAGEN™ somatic pipeline for calling small somatic variants from tumor samples, with or without paired normal samples. The DRAGEN somatic variant caller offers 1) a flexible architecture that can be used on a wide array of somatic use cases; 2) built-in noise models enabling robustness against various sources of noise artifacts (mapping, genome context, or sample specific); 3) performance of joint analysis of tumor and normal samples in the case of a tumor-normal workflow yielding improved accuracy; 4) benefits from FPGA acceleration for efficient run time. We demonstrate the speed and accuracy of the DRAGEN tumor-normal pipeline across a range of whole genome sequencing (WGS) datasets and compare against third party tools such as Mutect2/GATK4 [1] and Strelka2 [2]. DRAGEN secondary analysis outperforms all other tools with its ability to complete a 110x/40x T/N whole-genome analysis in less than two hours. It offers exceptional accuracy, with higher sensitivity and precision than third party tools. We also show that the DRAGEN T/N workflow supports analysis of liquid and late-stage solid tumors by tolerating tumor-in-normal (TiN) contamination.