Assessing non-LUS stutter in DNA sequence data

Forensic DNA analysis is among the most well-recognized and well-developed forensic disciplines. The field's use of DNA markers known as short tandem repeats (STRs) offer a robust means of discriminating individuals while also introducing challenges to the analysis. One of these challenges, stutter, is the result of a non-biological artifact introduced during PCR. The formation and amplification of these stutter products can occur at rates as high as 15-20% of the parent allele. The challenge inherent in this process is differentiating stutter artifacts from true alleles, particularly in the presence of a minor contributor. Traditionally, DNA profiles are obtained using capillary electrophoresis (CE), where amplified DNA fragments are separated by size, not sequence, and the identification of stutter is performed on a locus-specific level. The use of CE-based fragment data rather than sequence-based data, has limited the community's understanding of the precise behavior of stutter. Massively parallel sequencing (MPS) data provides an opportunity to better characterize stutter, permitting a more accurate means of detecting both size- or longest uninterrupted stretch (LUS)-based stutter but also allele and motif-specific stutter characteristics. This study sheds light on the value of characterizing motif- and allele-specific stutter, including non-LUS stutter, when using MPS methods. Analysis and characterization of stutter sequences was performed using data generated from 539 samples amplified with the ForenSeq and PowerSeq 46GY library preparation kit and sequenced on the Illumina MiSeq FGx. Assessment of non-LUS stutter begins with calculating stutter rates for all potential stutter products at a given locus (and allele), additionally, the occurrence of these discrete stutter products were quantified. Results show that although the LUS sequence stutters at a higher rate than non-LUS motifs, the non-LUS stutter products do occur at detectable levels and potentially influence sequence-based mixture analysis. The data indicate that the stutter from one motif or allele can be distinguished from another motif or allele based on their unique stutter rates; however, the number of stutter products from each motif or allele may similarly make up the overall pool of stutter products. Motif- and allele-specific stutter models provide the most comprehensive analysis of sequence stutter rates and provide the ability to differentiate stutter sequences more accurately from true allele stutter. This information provides a foundation for including the characterization of non-LUS stutter products when analyzing DNA profiles, specifically mixtures with potential low-level contributors.