Comparative evaluation of SNP sequencing workflows for identification of the missing in Vietnam

Hundreds of thousands of remains are still unidentified following the 1955-1975 conflict in Vietnam. DNA identification is increasingly challenging due to extreme degradation of remaining DNA, a large amount of contaminating microbial DNA, and a growing reliance on distant relatives to provide reference samples for comparison. Although mitochondrial DNA is easier to recover than nuclear DNA, by itself it does not have the discriminatory power needed for large-scale identifications. Standard nuclear short tandem repeat (STR) profiling is not an option due to degradation. Instead, massively parallel sequencing (MPS) and its ability to target thousands of single nucleotide polymorphisms (SNPs) must be considered. This paper presents a direct comparison of three targeted SNP sequencing workflows for the identification of highly degraded Vietnamese skeletal remains. Two pre-existing forensic identification mid-density SNP panels were selected for evaluation: the MPSplex panel and the FORCE panel. Both panels are available with library preparation workflows that have the potential of targeting and enriching SNPs from fragmented DNA: QIAseq single primer extension (FORCE and MPSplex panel) and hybridization capture (FORCE panel). Fifteen bone samples from fifteen sets of unidentified Vietnamese skeletal remains of varying quality were processed using three workflows (MPSplex and FORCE using a modified QIAseq protocol and FORCE using hybridization capture) to compare their performance. Family reference genotypes were simulated to estimate the theoretical identification power and performance of the workflows for each sample. All three workflows resulted in SNP profiles that would theoretically support identifications using at least a single first degree relative and up to a single fourth degree relative, from most of the 15 bone samples. The FORCE-Capture workflow had the highest performance based on SNP recovery and theoretical identification power, while FORCE-QIAseq had the lowest. The MPSplex-QIAseq workflow performed similarly to FORCE-Capture with respect to the percentage of targeted SNPs recovered per sample, however, the larger number of SNPs targeted by the FORCE panel appears to outweigh the advantage of the tri-allelic SNPs targeted by the MPSplex-QIAseq workflow. These findings demonstrate that the targeted sequencing of SNPs when combined with optimised, sensitive library preparation techniques, offers a viable path forward in the identification of the missing in Vietnam.