We evaluated the sensitivity of the methods with regard to allele fraction and tumor sequencing depth using the virtual tumor (Fig. 4a) and downsampling (Supplementary Fig. 6) approaches, and observed a sharp distinction in sensitivity, particularly at lower allele fractions. We analyzed data for 30x sequence coverage. In the standard configurations, all methods show ≥ 99.3% sensitivity for mutations at an allele fraction of 0.4. However, in the HC configurations, MuTect, JointSNVMix and Strelka remain sensitive, 98.8%, 96.6% and 98.5% respectively, whereas SomaticSniper drops to 91.5%. At an allele fraction of 0.1, MuTect HC can detect more than half of the mutations (53.2%), whereas Strelka HC detects only 29.7%, JointSNVMix HC drops to 16.8% and SomaticSniper HC falls to 7.4%. At an even lower allele fraction of 0.05, MuTect HC has 16.0% sensitivity but can be increased to 51.9% with 60x coverage. By contrast, JointSNVMix HC and SomaticSniper HC have a sensitivity of ≤ 2.0%, and the sensitivity does not increase appreciably with tumor sequencing depth. Strelka HC detects just 4.6% of the events at 30x and only increases to
