Genomic approaches to forensic DNA analysis

Genetic technologies for forensic identification have become highly sensitive and discriminating in recent years. Coupled with the implementation of genetic databases, forensic DNA analysis has enormously aided law enforcement efforts to identify criminals and connect unsolved cases. But some cases challenge the limits of current methods and technologies. Given my experience with forensic DNA analysis, I sought to apply genomic approaches to address two challenging aspects of forensic genetics: familial searching of DNA databases and interpretation of mixed DNA samples.;Familial searching can enhance the power of DNA databases by revealing profiles that are genetically similar to an unknown evidence profile. This information can provide investigative leads to find a close relative of a putative perpetrator. Working with large DNA databases, I evaluated the sensitivity and specificity of familial searching with 10--15 autosomal short tandem repeat (STR) markers and the increased discriminatory power of additional genetic information. For profiles based on 10--15 autosomal STRs, the number of false positive leads generated by a familial search was very large and increased with the size of the database and the distance of the relationship evaluated. Adding many more autosomal STRs, and especially including Y-chromosome STRs and mitochondrial DNA sequences, produced sensitive and specific familial searches to identify parent-offspring, full sibling, and half-sibling (or equivalent) relationships.;A challenge in the interpretation of DNA mixtures is to distinguish allelic peaks reflecting contributor genotypes from amplification artifacts. Two parameters to evaluate are thresholds for stutter and heterozygote balance. A recommended practice is to apply uniform thresholds to all loci, including 15% stutter and 60% heterozygote balance. To evaluate whether these thresholds might be too conservative, leading to loss of true allelic information, I evaluated stutter peaks and heterozygous genotypes from single-source samples. In subsequent blind tests of two-person mixtures, empirical thresholds based on allele-specific stutter significantly improved identification of minor contributor alleles without introducing false positive genotype assignments. A heterozygote balance threshold of 50% captured more true heterozygous alleles. Allele-specific stutter thresholds and a 50% threshold for heterozygote balance could be incorporated into the recommended mixture interpretation strategy.