NCJ Number
192036
Journal
Journal of Forensic Sciences Volume: 46 Issue: 6 Dated: November 2001 Pages: 1372-1378
Date Published
November 2001
Length
7 pages
Annotation
This study introduced a new automated approach for resolving forensic DNA mixtures.
Abstract
With the advent of PCR-based STR typing systems, mixed samples can be separated into their individual DNA profiles. Quantitative peak information can help in this analysis; however, despite such advances, forensic mixture analysis still remains a laborious art, with the high cost and effort often precluding timely reporting. The underlying PCR amplification step, as well as the fluorescent detection step, show a quantitatively linear response in the presence of DNA mixtures. This suggests the use of linear algebraic models to explain mixture problems and compute their solutions. This study introduced linear mixture analysis (LMA), a straightforward mathematical method for resolving DNA mixture problems. The underlying linear mathematics permits rapid and robust solutions on real quantitative data. LMA uses all the data in a single combined computation, which contributes to its robustness and accuracy (the method is unlikely to find an incorrect solution). Moreover, heuristic algorithms based on LMA have built-in approaches for determining error, identifying suspect loci, and establishing confidence. 3 tables, 2 figures, and 9 references