November 19, 2012
Speaker: Dr. Catherine Grgciak, School of Medicine Biomedical Forensic Sciences, Boston University
Title: Complex, Low-Template DNA Mixture Interpretation of Forensically Relevant STR Loci
Abstract: Accurate interpretation of DNA evidence depends upon the ability of a forensic practitioner to compare the DNA profile obtained from an item of evidence to the DNA profile of a standard. If a suspect is deemed a possible contributor to an item of evidence, then a statistic – which is an assessment of the significance of an inclusion – must be reported to the court. The goal of this presentation is to introduce the attendees to the difficulties associated with processing, analyzing and interpreting complex, low-level DNA samples and to introduce methods designed to overcome these challenges.
Currently, the ‘weight of a match’ is expressed in terms of the Random Match Probability (RMP), the Likelihood Ratio (LR) or the Combined Probability of Exclusion (CPE). In order to calculate this match-statistic the number of contributors must be known. If the item of evidence contains a mixture of cells from greater than two donors, or there are a limited number of cells, accurately determining the true number of contributors to a biological stain using traditional methods becomes difficult. Furthermore, traditional methods do not establish the actual number of contributors; instead the minimum number of contributors is determined and is based on counting the number of observed peaks. This counting method is prone to error as amplifying low levels of DNA is predisposed to stochastic effects which are observed in the form of allele drop-out. To complicate matters, short tandem repeats, have limited variability within the population, resulting in a significant level of allele overlap in samples with a large number of contributors. Recent efforts to accurately determine the number of contributors using likelihood estimates have been proposed, but do not take into account the probability that an allele may have ‘dropped-out’ or ‘dropped-in’. An approach which calculates an a posteriori probability (APP) of the number of contributors to a stain based on the genotyping results will be introduced. The APP is the probability that the stain came from a certain number of contributors given what is observed during genotyping. If it is strongly peaked, i.e. the APP says that there is a particular number of contributors that is highly likely and all others are highly unlikely, then the APP tells us the number of contributors that gave rise to the stain. If not, the APP will give the range in which the number of contributors is overwhelmingly likely to lie.
As previously stated, the Likelihood Ratio approach is one of the traditional methods used to assess the strength of DNA evidence comparisons. It assesses the DNA evidence probabilistically and compares the hypotheses of the prosecution and defence and utilizes assumptions which must be made prior to analysis. However, the signal obtained from biological evidence may be so complex that the signal obtained from the person-of-interest may not be distinguishable from baseline noise, PCR/instrument artifact and multiple ‘other’ contributors. This presentation will introduce preliminary results that show the risk associated with interpreting low-level DNA mixtures. Increased risk with decreased template levels indicates that current methods do not completely fulfill the needs of today’s forensic DNA laboratories. A generalized interpretation scheme in which techniques borrowed from the digital communication field will be introduced as a possible alternative. By systematic comparisons of the crime scene profile to those from individuals of interest, it may be possible to include or exclude individuals as potential contributors without relying on relative comparisons of hypotheses.
