We can summarize the tradeoff between sensitivity and specificity for each of the methods using a ROC curve, which depends on the sequencing depths in the tumor and normal and the mutation allele fraction. Figure 4d gives an example using tumor depth of 30x, normal depth of 30x and allele fraction of 0.1, showing that MuTect is a generally more sensitive for a given specificity and also has a much smaller decrease in sensitivity for a similar increase in specificity gained by the HC configuration (dashed lines).
