An Assessment of Probabilistic Approaches to mtDNA Mixture Interpretation

Massively parallel sequencing (MPS) presents an opportunity to extend forensic mitochondrial (mt)DNA typing capabilities beyond the current standard. A high-depth MPS approach allows for the discrete resolution of low-level mixed nucleotide positions. A recognized challenge when performing mtDNA analysis is the inability to deconvolute mixtures. Mixtures have always been a part of forensic casework and have become more common in forensic investigations, with a retrospective study evaluating 691 casework hair samples indicating that about 9% of the samples were observed to have a mixture of mtDNA sequences. When samples with observations of mixed base calls are encountered, most forensic laboratories do not interpret the results and categorize the findings as inconclusive for reporting purposes. In order to retain the evidentiary value of these mixed samples, forensic laboratories need the ability to deconvolute mixed sequences, which will require software and bioinformatic solutions. The goals of this research are to 1) combine and assess two existing software tools, MixtureAce and Mixemt, to deconvolute mtDNA mixtures, 2) create a dataset of in-silica MPS mixtures from whole mitogenome haplotypes consisting of two- and three-person contributors at different dilution ratios to test the combined tools, and 3) generate biological mtDNA mixtures of two- and three-person contributors at various dilution ratios using two commercial amplification kits designed for mtDNA analysis on a MiSeq to further validate the software for use in forensic laboratories. The last goal will allow for an assessment of how different target amplicon strategies impact the mixture deconvolution process. Testing of the probabilistic pipeline will target a range of mixture characteristics. Mixtures with different haplotypes and disparate dilution ratios are generally easier to resolve and will serve as controls, while more similar haplotypes and equal dilution ratios will be used to evaluate the ability of the software to deconvolute more complex mixtures. In addition, a set of mixtures will include samples with known heteroplasmy to assess its impact on interpretation. The outcomes of this work will help to increase the effectiveness of forensic mtDNA analysis, decrease the time and cost of data analysis compared to current standard practices, allow for separation of components of a mixture, and will include quantitative measures and statistical evaluation of forensic evidence. Each of these outcomes will address a specific interest of the criminal justice system and NIJ. Note: This project contains a research and/or development component, as defined in applicable law, and complies with Part 200 Uniform Requirements - 2 CFR 200.210(a)(14). CA/NCF