Development of computational methods for genetic identification and kinship analysis of forensic samples

Advancement in next-generation sequencing technologies have enabled the recovery of genetic data from minute, contaminated, or degraded samples, removing roadblocks to many forensic investigations. Nevertheless, conventional approaches for identity analysis, often relying on PCR-based determination of short tandem repeats (STRs), fail when poor-quality DNA prevents full genotype calls at the sites used for comparison. Furthermore, currently there is no efficient method for matching limited sequence information against high-quality single-nucleotide polymorphism (SNP) profiles, although these data can be respectively generated from crime-scene specimens and persons of interest. To address these challenges, I propose the development of reliable computational methods for genetic identification and kinship analysis using low-coverage sequencing data from forensic samples. For my first aim, I have implemented a program, IBDGem, that utilizes identity-by-descent (IBD) models of relatedness to compare low-coverage sequencing data from an unknown sample and genotype data from a known individual. I have found that IBDGem can make accurate identity detections at coverages as low as 0.01x, and is robust to the choice of population allele frequencies. My next two aims will focus on improving this method by taking into account linkage disequilibrium and incorporating genetic linkage maps to optimize kinship estimation. I will evaluate the program using down-sampled sequence and genotype data from 1000 Genomes, as well as low-coverage DNA data from rootless hair samples. I have made IBDGem freely available for non-commercial applications on Github, and the program is already in practical use. This technology also has the potential to be licensed by companies that assist law enforcement in forensic investigations. One of the key challenges to the wide adoption of this tool would be its incompatibility with CODIS STR profiles. Nevertheless, with the proliferation of direct-to-consumer genetic testing and use of forensic genetic genealogy investigative techniques, IBDGem still offers a powerful and practical solution for forensic genetic identification. I will describe the technical details of the software, together with the results obtained from this proposal in peer-reviewed publications and reports submitted to the NIJ, so that they become accessible to a broader audience. CA/NCF