Exploring the probative value of mixed DNA profiles. (July 2019)

• Record Type: Journal Article
• Authors: Kruijver, Maarten; Bright, Jo-Anne; Kelly, Hannah; Buckleton, John
• Is Part Of: Forensic science international. Volume 41(2019)
• Page Start: 1
• Page End: 10
• Publication Date: 2019-07
• Subjects: DNA mixture interpretation -- Probabilistic genotyping -- Adventitious match -- Database search; Forensic genetics -- Periodicals; Génétique légale -- Périodiques; Forensic genetics; Electronic journals; Periodicals; 614.1
• DOI: 10.1016/j.fsigen.2019.03.010
• Languages: English
• ISSNs: 1872-4973

Abstract

Highlights: The probative value of a (mixed) profile can be estimated in the absence of a POI. An expected range of likelihood ratios is obtained for the unknown contributors. The risk of an adventitious match is obtained on a per case basis. Expected evidential efficiency summarises the probative value. NIST MIX13 Cases 3 and 4 and an example from the PROVEDIt sets are reworked. Abstract: Probabilistic genotyping typically proceeds by first deconvoluting a mixture into separate components and then computing a likelihood ratio for a potential contributor. The typical range of likelihood ratios for contributors and unrelated profiles depends, to a large extent, on how well the mixture is resolved. This in turn depends on the quality and complexity of the sample. Some samples are highly discriminatory while others are likely to yield only inconclusive results. Knowing the power of discrimination is helpful when deciding on whether or not to proceed with investigating a case or to do potential rework and for deciding on database inclusion. We present a method for exploring the range of likelihood ratios expected under a proposition of choice. We describe the results of several case scenarios exploring questions such as: 'Is the profile's minor contributor likely to yield a large likelihood ratio?', 'Is it feasible to find a donor among a large database of unrelated profiles?', or 'What is the probability that a brother of a true donor yields a large likelihood ratio?'. We demonstrate the efficiency and speed of the method for a number of examples.