Single-cell Likelihood Ratios are Highly Informative and Robust Across Multifarious Mixture Complexities

Single-cell Likelihood Ratios are Highly Informative and Robust Across Multifarious Mixture Complexities

 

Single-cell Likelihood Ratios are Highly Informative and Robust Across Multifarious Mixture Complexities

Catherine M. Grgicak*, Desmond S. Lun, Madison M. Mulcahy, and Nidhi Sheth | Rutgers University–Camden Ken R. Duffy and Leah O’Donnell | Maynooth University Abstract: Traditional mixture interpretation works within a likelihood ratio (LR) paradigm and requires assumptions on relatedness and whether a known contributor’s DNA is present in the data. In the bulk pipeline, DNA is extracted using laboratory treatments that force volume fractionation between the extraction and polymerase chain reaction (PCR) steps. This fractionation leads to variation in the number of copies of each allele before PCR cycling starts, which drives peak height ratio differences between alleles within a locus and drop-out. The resultant bulk electropherogram (EPG) is, therefore, a mix of short tandem repeat (STR) peaks from possibly related individuals who may not have their full profiles represented in the data. It is for these reasons that multifarious number of contributors assignments and contexts are required to evaluate the evidence. As the number of reasonable propositions increases so do the cost burdens associated with training, proficiency testing, and research. Though technologies, such as sequencing, might improve the ability to resolve more alleles, full resolution can only be acquired through a single-cell approach that invokes a direct-to-PCR methodology. Thus, the researchers endeavor to develop a single-cell pipeline, complete with interpretation and capable of reporting the strength of the consistency between group(s) of single-cell data and a person of interest (POI). Despite its potential, single-cell analysis has not received widespread consideration within the forensic domain and therefore requires studies that affirm its relevance. To do this, this study must demonstrate the flexibility of this strategy by showing that for a set of mixtures containing two to five persons, the LR for a cluster of single-cell electropherograms (scEPGs), grouped by similarity, approaches that of the known genotype regardless of the number of individuals comprising the mixture. This study shall also present a treatment that demonstrates how the strength of consistency between scEPGs in a cluster and the POI can be combined to determine the strength of evidence for the entire mixture. The presentation shall continue by demonstrating two boundary cases: (1) where the LRs for a mixture consisting of 15 persons is determined and (2) where each cluster contains only one scEPG. In the first case, the mixture is tested against all 15 true contributors and finds that all the LRs approach those of the ground truth genotype. From the case where each cluster contains only one scEPG, this study found that a total peak height of 5,000 RFU, representing ca. 10 allele peaks, renders LR ≈ 106 and that over 80% of the cells exhibit at least this allele detection rate. The scLRs from this software application, named EESCIt for Evidentiary Evaluation of Single Cells, were perfectly repeatable and were obtained in less than 1 minute for any admixture. These results demonstrate the relevance, salience, and legitimacy of this strategy to the forensic domain, justifying further development.