STRmix™ put to the test: 300 000 non-contributor profiles compared to four-contributor DNA mixtures and the impact of replicates.

Probabilistic genotyping approaches are increasingly used for the interpretation of DNA mixtures. To explore the specificity of one of these systems (STRmix™), we conducted an extensive study using 24 complex mixtures: all were known or apparent 4-person mixtures with at least one contributor representing less than 20% of total DNA, and all mixtures had at least one contributor with suboptimal DNA quantity. Those mixtures were either generated in-house or from casework. All the mixtures were compared to 300,000 virtual non-contributors, resulting in a dataset of 7.2 million comparisons. The great majority of the non-contributor comparisons led to a LR lower than 1 for a specificity of 99.1%. The effect of using replicate amplifications to calculate the LR of non-contributors was also assessed as triplicates were used and led to an increased specificity of 99.8%. The very large extent of the analyzed data shows that STRmix™ has an excellent ability to discriminate non-contributors from complex DNA mixtures. Crown