Advanced statistical population genetics methods for forensic DNA identification.

As submitted by the proposer: Forensic DNA profiling has long relied on principles and methods from the science of human population genetics to provide models and tools for assisting in accurate statistical evaluation of genetic evidence. This project will bring transformative new techniques from population genetics to enhance and catalyze innovation in three aspects of DNA profiling. First, we will use new methods of genotype imputation relying on linkage disequilibrium to enable microsatellite genotypes of the CODIS system to be statistically predicted from genotypes obtained using single-nucleotide polymorphisms (SNPs). This work will determine the extent to which new SNP systems - which, in comparison to the CODIS microsatellite system, can offer advantages in cost, standardization, and information content for such challenges as relatedness profiling and DNA mixtures - can be produced in a manner that is statistically backward-compatible with large existing databases of microsatellite profiles. Second, we will use recently developed methods of ancestry inference to maximize the information that can be extracted from CODIS loci about the population of origin for sampled profiles. In relatedness profiling, in which a crime sample is examined for partial matches that could indicate close familial relationships between the crime sample and tested samples, misspecified population of origin increases the rate at which tested samples are erroneously identified as matches; we will use ancestry information to avoid population misspecification and thus to reduce false positives in relatedness profiling. Third, we will perform detailed assessments of the role of population of origin in DNA mixture profiling, in which crime samples contain mixtures of DNA from two or more contributors. We will additionally use ancestry inference to reduce false positives in mixture profiling. The project will be conducted using a sample of 994 individuals from 52 globally distributed populations, including existing SNP data and new CODIS profiles that we have generated. These new individual multilocus CODIS profiles will be publicly released as part of a dissemination plan that also includes publications in forensic science and genetics journals, presentations at forensic science conferences, and software. The work will facilitate linkages of population-genetic and forensic DNA research, and it includes training of postdoctoral scientists and PhD students on forensic problems. The project will both enhance the potential of the CODIS system for relatedness profiling and DNA mixtures, and, through cross-compatibility of SNPs and microsatellites, it will unleash new possibilities of improved forensic marker systems that can be statistically integrated with existing CODIS databases. ca/ncf