Direct deconvolution of unknown DNA profiles in a 2-person mixture for use in investigative forensic genetic...

We present an automated pipeline for the deconvolution of DNA mixtures using SNPs from the ForenSeq® Kintelligence® kit aimed at facilitating searching in forensic investigative genetic genealogy (FIGG) databases. This pipeline automates the entire process, from data ingestion to result generation, specifically tailored for sequencing result files from the Universal Analysis Software for ForenSeq Kintelligence. The pipeline determines the number of contributors in a DNA mixture and, for each unknown contributor in a two-person mixture, estimates their percent contribution, biological sex, and genotype across SNP loci. Additionally, it assigns probability values to the predicted DNA profiles, enhancing confidence in the deconvolution outcomes. The pipeline's efficacy was demonstrated through 10,000 in silico experiments and 55 empirical DNA mixtures, including a search for relatives in GEDmatch PRO™ using the deconvolved profiles, illustrating its high performance and feasibility. Importantly, deconvolved profiles produced from this tool were able to match to close relatives when queried against GEDmatch PRO, although many false positives were also recovered. With further development and optimization, this tool has the potential to address a significant gap in FIGG applications where mixed DNA samples have previously hindered the investigation of missing persons and criminal suspects.