Background/Objectives: Interpretation of mixture profiles generated from crime scene samples is an important element in forensic genetics. Here, a workflow for mixture deconvolution of sequenced microhaplotypes (MHs) and STRs using the probabilistic genotyping software MPSproto v0.9.7 was developed, and the performance of the two types of loci was compared. Methods: Sequencing data from a custom panel of 74 MHs (the MH-74 plex) and a commercial kit with 26 autosomal STRs (the ForenSeq™ DNA Signature Prep Kit) were used. Single-source profiles were computationally combined to create 360 two-person and 336 three-person mixtures using the Python script MixtureSimulator v1.0. Additionally, 72 real mixtures typed with the MH-74 plex and 18 real mixtures typed with the ForenSeq Kit from a previous study were deconvoluted using MPSproto. Results: The deconvoluted MH profiles were more complete and had fewer wrong genotype calls than the deconvoluted STR profiles. The contributor proportion estimates were more accurate for MH profiles than for STR profiles. Wrong genotype calls were mostly caused by locus and heterozygous imbalances, noise reads, or an inaccurate contributor proportion estimation. The latter was especially problematic in STR sequencing data, when two contributors contributed equally to the mixture. A total of 34,800 deconvolutions of the simulated mixtures were performed with two defined hypotheses: Hp, “The sample consists of DNA from one/two unknown contributor(s) and the suspect” and Hd, “The sample consists of DNA from two/three unknown individuals”. All true contributors were identified (LR > 1015 for MHs and LR > 109 for STRs) and all non-contributors excluded (LR < 10−6 for MHs and LR < 0.2 for STRs). Conclusions: In simulated and real mixtures, the MHs performed better than STRs.
Imputing single nucleotide polymorphisms (SNPs) to expand genotyped SNP panels is commonly applied to study genotype-phenotype correlations in medical and population genetics but is largely unexplored in forensics. Forensic DNA phenotyping, i.e., the SNP-based phenotype prediction, can greatly benefit from imputing missing prediction model SNPs. Unfortunately, most imputation studies investigate the performance of random SNPs with limited focus on phenotype-associated SNPs. Here, we compare the imputation accuracy of SNPs leveraged in phenotype prediction models, phenotype-associated SNPs and all SNPs and explore the performance of phenotype prediction model, HIrisPlex-S, using imputed datasets. We confirm that the number and selection of SNPs in the genotype dataset and minor allele frequencies (MAFs) are major drivers of imputation call and error rates. Phenotypic SNPs show higher imputation error rates than randomly selected SNPs due to MAF differences. Further, we provide novel insights into accurate phenotype prediction for dense SNP panels of high linkage, frequent phenotypes in the reference panel, and lenient imputation thresholds due to the detrimental effect of missing genotypes on trait prediction compared to imputation errors. Overall, we show the importance of imputation performance studies for SNPs of interest and the high applicability of SNP imputations prior to phenotype prediction.