This study evaluates the use of benchmarking for genotyping and imputation using degraded DNA for forensic applications across diverse populations.
In this study, researchers simulated sequencing data of varying qualities–coverage, fragment lengths, and deamination patterns–from forty individuals of diverse genetic ancestries. The researchers used this dataset to test the performance of commonly used genotype and imputation methods (SAMtools, GATK, ATLAS, Beagle, and GLIMPSE) on five different SNP panels (MPS-plex, FORCE, two extended kinship panels, and the Human Origins array) that are used for forensic and population genetics applications. The authors provide benchmarks and recommendations for analyzing degraded DNA from diverse populations with widely used genomic methods in forensic casework. The authors find that ATLAS outperforms GATK and SAMtools, achieving over 90% genotyping accuracy for the four largest SNP panels with coverages greater than 10X. For lower coverages, decreased concordance rates are correlated with increased rates of heterozygosity. For genome mapping and variant calling with degraded DNA, the study found that use of parameters and methods (such as ATLAS) developed for ancient DNA analysis provides a marked improvement over conventional standards used for next generation sequencing analysis. Genotype refinement and imputation improve the accuracy at lower coverages by leveraging population reference data. For all five SNP panels, the authors find that using a population reference panel representative of worldwide populations (e.g., the 1000 Genomes Project) results in increased genotype accuracies across genetic ancestries, compared to ancestry-matched population reference panels. Importantly, the authors find that the low SNP density of commonly used forensics SNP panels can impact the reliability and performance of genotype refinement and imputation. This highlights a critical trade-off between enhancing privacy by using panels with fewer SNPs and maintaining the effectiveness of genomic tools. (Published Abstract Provided)