NCJ Number
216799
Journal
Journal of Forensic Sciences Volume: 51 Issue: 6 Dated: November 2006 Pages: 1284-1297
Date Published
November 2006
Length
14 pages
Annotation
This paper presents a least-square deconvolution (LSD) framework for interpreting short-tandem-repeat DNA mixture samples contributed by two persons, using the quantitative allele peak data information.
Abstract
Results from studies that use simulated data as well as from forensic case data show that LSD consistently produces the correct resolution of profiles, provided no severe allele degradation or peak-height saturation from overload DNA samples exists among the given peak data of a locus. Some limitations are imposed by the inherent mathematical properties linked with similar solutions for two-allele loci. Using the suggested LSD interpretation guidelines may address some of these limitations. The LSD framework for mixtures with proportions close to 1:1 yields less certain results for four-allele and three-allele loci due to possible heterozygous peak imbalance (four-allele loci); this masks the slight difference in the mass proportion ratio. Degeneracy (for three-allele loci) occurs when two combinations have the same pattern of relative peak heights. The use of a known contributor profile aids in the final determination of resolution when ambiguities occur. An expert system for interpreting the LSD mathematical results is currently being developed by the authors. The proposed framework applies the least-square principles to calculate a best-fit mass proportion ratio for all possible genotype combination cases at each locus, independent of all other loci. Based on the ranked best-fit mass proportion ratios and the relative error residual ratios, a composite profile for each contributor can be developed from the LSD mathematical results by using the heuristic interpretation guidelines explained in this paper. 20 tables and 18 references