Results, methodology, and implications for criminal justice are reported for a research project that sought to increase understanding of complex forensic DNA interpretation and to continue the development of a novel approach in interpreting low-template DNA samples that contain many contributors.
The project succeeded in upgrading a continuous method to compute the LR and LR distribution based on modeling the peak heights by simulation of genotypes based on allele frequencies, while modifying a previously developed full mechanistic model of the forensic laboratory process for purposes of forensic validation. CEESIt - which is built on a continuous mixture interpretation model that incorporates noise, stutter, stochastic PCR effect, and random contributor levels - behaved as expected. It showed that if the LR for a true contributor is generally larger and the probability that the LR determined for a non-contributor is greater than one is reduced when more information is imported into CEESIt. In addition, the project implemented a structured software-testing process, which suggested that CEESIt v 3.1 is fit for its intended use and is robust. The description of project design and methods addresses the generation of 2,400 well-defined DNA mixture profiles with expanded STR amplification kits, refinement of the models in CEESIt and model degradation, and the development of a user-friendly GUI. Data analysis is also described. 4 figures, 3 tables, and listings of 10 project scholarly products and 6 references