The discriminatory power of STRmix illustrated by ROC curves

The use of probabilistic genotyping methods has seen a significant uptake in recent years throughout the world. There is a continuing need for empirical validation of such methods in contexts involving different wet chemistry conditions and various types of (mixed) samples. We have published a large scale empirical validation study in response to the 2016 PCAST report addressing, specifically, the issue that some sample categories were perceived to have received little, to no, attention in the empirical validation literature. More recently, the use of receiver operating characteristic (ROC) analysis was suggested as an important part of such validation exercises. We present the results of ROC analysis for a previously published study. The ROC curves demonstrate the great discriminatory power of DNA demonstrated using the probabilistic genotyping software STRmix™.