An investigation of the robustness of the consensus method of interpreting low-template DNA profiles.

Forensic STR profiles generated from low-template DNA samples are more noticeably subject to effects such as allele dropout, contamination with spurious alleles ('drop-in') and proportionally larger stutter. The profiles obtained are frequently partial, and are challenging to interpret. Specifically, interpretation guidelines which are used when the template DNA is in the optimal range for the STR test kit in use must be adapted to the low-template regime. A commonly used approach to such modified interpretation is known as the 'consensus' or 'biological' method, and relies on replication to achieve reliable results. We have carried out a study to assess the robustness of the consensus method as applied to SGM Plus(®) STR profiles obtained after applying a set of post-PCR purification methods together known as DNA SenCE, and report the results here. Multiple repeat analysis of DNA at five template quantities (ranging between 5pg and 100pg) and from five single donors, was carried out, and the resulting profiles were used to produce consensus profiles using several different evaluation criteria. Our aim was to determine whether the consensus profiles produced are conservative, that is, that the alleles reported are associated with the donor and that drop-in is reduced or eliminated. To this end, the alleles in the consensus profiles were compared with those of the donors, and the degree of concordance determined. The results suggest that increasingly stringent requirements for the number of times an allele must be observed in a set of repeat runs do, as expected, reduce the effect of drop-in, but also reduce the evidential value of the consensus profiles. However, the former is reduced to a much greater extent than the latter, resulting in a relative increase in profile information content versus drop-in peak risk with increased stringency. We also found that approximately half of the non-donor peaks appearing in consensus profiles were in -4 stutter positions for donor alleles present in the same profile, suggesting that many of these so-called drop-in alleles are, in fact, large stutter peaks rather than 'true' drop-in. Nevertheless, the appearance of non-donor peaks in a profile, including what are assumed to be oversized stutter peaks, appears to be an essentially random event.