The Qiagen Investigator® HDplex Kit provides valuable additional information to assess complex kinship scenarios, by supplementing the set of forensic standard short tandem repeat (STR) loci [1]. We genotyped 1198 Swiss individuals from a previous population study for those nine additional loci.

DNA extractions were prepared as described [2]. Multiplex PCR was performed in a reduced reaction volume of 12.5 μL. Capillary electrophoresis was run on a 3500xl genetic analyzer (ThermoFisher, USA) and data interpretation carried out with Genemapper® ID-X, v1.4 (Thermo Fisher, US). Details on the population structure can be found in [2]. Allele frequencies, forensic parameters, and test for Hardy–Weinberg equilibrium (HWE) with 10,000 permutations were calculated using STRAF [3], excluding two genotypes with three-allelic patterns. Results are listed in Table S1. None of the nine additional loci shows significant deviation from HWE. Allele frequencies are also available as Familias [4, 5] input file for all 32 loci (Table S2). Pairwise FST values for the six regional sub-populations defined in [2] revealed no intra-national differences, as expected (Table S3).

All genotype data for the three loci D12S391, D18S51, and SE33 are concordant with the data previously generated with Promega PowerPlex® Fusion 6C [2]. One sample showed a triplet in D6S474, another one in D10S2325. Five samples showed an almost complete allele dropout in marker D2S1360. This partial dropout has previously been observed and has been shown to be due to a SNP in the primer binding site [6]. Interestingly, in our dataset, all partial dropouts concerned allele 21, suggesting an association in the studied population of this allele with the SNP in the flanking region. We also detected a couple of off ladder alleles. All variants are listed in Table S4.

For marker D21S2055, heterozygote balance was usually below 60% if one allele consisted of less than 30 repeats and the other one of more than 30 repeats, limiting the utility of this marker for trace analysis and particularly DNA mixtures. However, there is no clear-cut allele length boundary for this phenomenon. We also observed pronounced peak imbalance between a couple of other pairs of alleles, e.g., 17.1/29 or 31/33. A similar observation has already been made by Tillmar et al. [7]. This suggests an association of a structural difference in the flanking regions with the STR repeat length, which might be worth a future investigation by sequencing.

Expanding the STR dataset increases the number of syntenic loci. Since ignoring linkage between loci might have an impact on certain kinship scenarios [8], we list the information for relevant syntenic loci, including loci distances based on Phillips et al. [9], in Table S5. We checked for linkage disequilibrium with the genepop R package [10, 11] by performing an exact test with 50,000 iterations and 1000 batches. For the complete dataset of 32 loci, we could not detect any significant linkage disequilibrium after Bonferroni correction (see Table S6).

The present dataset has passed the quality control for STRidER [12] and obtained the STRidER accession number STR000368.