Advancing Probabilistic Approaches to Interpreting Low-template DNA Profiles and Mixtures: Developing Theory, Implementing Practice

As submitted by the proposer: Samples containing low-levels of DNA and/or mixtures of DNA from multiple individuals are routinely encountered in forensic DNA casework. Such samples are challenging to interpret because of the inherent uncertainty in determining the genotypes of the contributors to the evidence profile. Such uncertainty must be taken into account when assessing the weight of such complex DNA profiles. Probabilistic approaches, all of which employ a likelihood ratio (LR) framework, have already been established to aide in the interpretation of such profiles. However, the performance of any of these approaches has yet to be rigorously evaluated. Further, the forensic DNA community is in need of freely available, user-friendly tools to implement a probabilistic approach to the determination of the weight of evidence. The proposed research will fill both of these voids by examining the performance of various probabilistic approaches as applied to complex DNA profiles, increasing our understanding of the capabilities and limitations of these approaches, and improving and extending an existing freely available program that can assist in understanding the weight of evidence of these complex DNA profiles. Specifically, we will: 1) Generate a set of low-template mixtures. This dataset will be used in subsequent parts of the project and will be made available to the community for evaluation of different statistical approaches. 2) Evaluate the performance of estimates of drop-out probabilities in mixed samples. Approaches to calculating LRs either require, or may be improved by, an estimate of the drop-out probability. We have previously shown how such estimates can be derived from single-source samples, how the estimates compare to the true (P(DO)s, and how LRs computed with each compare. Here we will evaluate how those estimators perform when applied to mixtures. 3) Compute and evaluate results of LRs for complex mixtures of 2, 3 and 4 contributors. The purpose of this project is to assess the information content of such complex mixtures, and the extent of support for a proposition that a specific individual is a contributor to the mixture. If such mixtures contain little information, then this might suggest changes to laboratory policy for interpreting such samples. We will employ both simulated and real databases to explore the result of performing LRs conditioned on a wide range of known non-contributors. These experiments will greatly assist in understanding the meaning of weaker LRs obtained for suspected contributors. 4) Evaluate the role of stutter in mixture interpretation. In particular, we will explore how stutter and a minor contributor with an allele in the stutter position of a major contributor peak combine to determine the final peak height. 5) Extend our user-friendly, freely-available software program, Lab Retriever, to help analysts perform complex LR calculations for low-template DNA mixtures. The end result of this work will be increased knowledge on how to interpret complex DNA profiles and new tools that will aide forensic DNA analysts in accurately and efficiently assessing the weight of such evidence.
ca/ncf