Interpreting forensic DNA profiling evidence without specifying the number of contributors

• Published in: Forensic science international : genetics Vol. 13; pp. 269 - 280
• Main Authors: Taylor, Duncan, Bright, Jo-Anne, Buckleton, John
• Format: Journal Article
• Language: English
• Published: Netherlands 01.11.2014
• Subjects: DNA Fingerprinting - methods; Genotype; Humans; Likelihood Functions; Markov Chains; Models, Genetic; Monte Carlo Method; Pathology; Reproducibility of Results; MCMC; STRmix; Mixtures; DNA profile interpretation; Number of contributors; Continuous model
• ISSN: 1872-4973; 1878-0326
• DOI: 10.1016/j.fsigen.2014.08.014

Abstract DNA profile interpretation has benefitted from recent improvements that use semi-continuous or fully continuous methods to interpret information within an electropherogram. These methods are likelihood ratio based and currently require that a number of contributors be assigned prior to analysis. Often there is ambiguity in the choice of number of contributors, and an analyst is left with the task of determining what they believe to be the most probable number. The choice can be particularly important when the difference between two possible contributor numbers means the difference between excluding a person of interest as being a possible contributor, and producing a statistic that favours their inclusion. Presenting both options in a court of law places the decision with the court. We demonstrate here an MCMC method of correctly weighting analyses of DNA profile data spanning a range of contributors. We explore the theoretical behaviour of such a weight and demonstrate these theories using practical examples. We also highlight the issues with omitting this weight term from the LR calculation when considering different numbers of contributors in the one calculation.