Concordance and characterization of massively parallel sequencing at 58 STRs in a Tibetan population.

BACKGROUND: Massively parallel sequencing (MPS) is a promising supplementary method for forensic casework in short tandem repeats (STRs) genotyping, owing to several advantageous features in comparison to traditional capillary electrophoresis (CE). However, the application of MPS in casework requires accessible datasets from the worldwide population to enrich the allele frequencies of sequence-based STR genotypes.
METHODS: In this study, we report the characterization of sequence-based allele frequencies of 58 STRs from a Tibetan population comprising 120 unrelated individuals using the ForenSeq™ DNA Signature Prep Kit. A concordance study evaluating MPS and CE allele data was performed to ensure that MPS is compatible with current CE-based forensic databases. The diversity of observed alleles, allele frequencies, and forensic parameters per locus by length (LB), sequence without flanking region (RSB), and sequence with flanking region (FSB) were analyzed and compared.
RESULTS: The concordance study demonstrated a concordance rate exceeding 99%. The combined random match probability (RMP) for the 26 A-STRs was 2.04 × 10-29 , 1.93 × 10-31 , and 9.56 × 10-33 for LB, RSB, and FSB, respectively. Similar trends were observed in other forensic parameters resulting from the increase in the number of unique alleles available. A total of 111 and 113 unique haplotypes in the Y-STR loci were observed when using length-based and sequence-based alleles, respectively. In addition, we identified 35 novel alleles at 25 loci and 25 polymorphisms in the flanking regions at 17 STRs.
CONCLUSIONS: Our data suggest that MPS- and CE-derived alleles are compatible. MPS-based analysis of the STR data substantially increased the allele diversity and improved the forensic parameters, which clearly demonstrated the advantages of MPS in comparison to CE. With more pooled data and larger-scale validation, MPS could play a valuable role in forensic genetics and might be an additional tool for routine casework.

INTRODUCTION

Short tandem repeats (STRs) are known to be ubiquitous across the human genome, and present with sufficient variability to allow for the identification of individuals, making them ideal in forensic genetic applications (Butler, 2005; Jobling & Gill, 2004). STRs are routinely analyzed using capillary electrophoresis (CE), which is considered the gold standard for forensic genetics and has been widely recognized in criminal investigations and prosecutions for over two decades (Butler, 2015; Butler et al., 2004; Thompson et al., 2012). However, the CE method only detects amplicon length while overlooking potentially informative sequence variation. The high polymorphism rates of STRs are underutilized.

Massively parallel sequencing (MPS), also known as next‐generation sequencing (NGS), has been shown to have potential in forensic genetics over the past few years (Alvarez‐Cubero et al., 2017; Aly & Sabri, 2015; Børsting & Morling, 2015; Bruijns et al., 2018). The potentially informative sequence variations in STRs (both in the repeat and in the flanking regions) can be evaluated by MPS, broadening STR diversity and increasing the discrimination powers of analytical tests (Barrio et al., 2019; Churchill et al., 2017; Delest et al., 2020; Gettings et al., 2015, 2016; Hussing et al., 2019; Jäger et al., 2017; Khubrani et al., 2019; Kim et al., 2016, 2018; Novroski et al., 2016; Peng et al., 2020; Phillips, Devesse, et al., 2018; Phillips, Gettings, et al., 2018; Wang et al., 2020; Wendt et al., 2017). In addition, MPS allows for the simultaneous analysis of significantly more loci than CE as it is not limited by restrictions in size‐based separation or fluorescence dye detection (Li et al., 2017). The smaller amplicon sizes in MPS may also improve the analysis of challenging or degraded DNA samples (Elwick et al., 2019; Fattorini et al., 2017; Kuffel et al., 2020). Given these advantages MPS is a promising supplementary method for forensic casework.

The application of MPS in casework requires accessible datasets from the global population designed to enrich the allele frequencies of sequence‐based STR genotypes, as recommended by the International Society for Forensic Genetics (ISFG) (Parson et al., 2016; Phillips, Devesse, et al., 2018; Phillips, Gettings, et al., 2018). Several MPS‐STR population datasets have been reported in recent years (Churchill et al., 2017; Delest et al., 2020; Hussing et al., 2019; Khubrani et al., 2019; Kim et al., 2016, 2018; Novroski et al., 2016; Peng et al., 2020; Phillips, Devesse, et al., 2018; Phillips, Gettings, et al., 2018; Wang et al., 2020; Wendt et al., 2017). Tibetans are one of China's 56 ethnic groups and are indigenous to the Qinghai‐Tibet Plateau. In China, Tibetans are primarily distributed across the Tibet Autonomous Region in the Qinghai, western Sichuan Province, Diqing in Yunnan, and Gannan in Gansu. Additionally, some Tibetans live in India, Bhutan, the United States, Canada, Europe, Australia, and other parts of the world. In this study, we report the characterization of sequence‐based allele frequencies from 58 STRs in a Tibetan population comprising 120 unrelated individuals using the ForenSeq DNA Signature Prep Kit (Verogen, San Diego, CA, USA), an MPS panel validated by several researchers and laboratories worldwide (Guo et al., 2017; Köcher et al., 2018; Wu et al., 2019). Notably, the genotypes obtained using MPS must be consistent with those obtained by CE to ensure that these data are compatible with current forensic databases (Devesse et al., 2018, 2020). Therefore, prior to the characterization of the sequence variation and the allele frequencies, a concordance study between the two methods was performed in our study.

MATERIALS AND METHODS

DNA sampling, extraction, and quantification

Peripheral blood samples were collected using FTA cards from 120 unrelated male individuals who claimed to be indigenous Tibetans residing in Lhasa, the capital of Tibet, who could track their heritage by at least three generations. All individuals provided written informed consent and genomic DNA was extracted using the BioRobotEZ1 Advanced XL and EZ1 DNA Investigator kits (Qiagen) according to the manufacturer's instructions. DNA was then quantified using a Qubit 2.0 Fluorometer and a Qubit dsDNA HS Assay Kit (Thermo Fisher). This study was approved by the Fudan University ethics committee (2020016).

Library preparation and sequencing

Libraries were prepared using the ForenSeqTM DNA Signature Prep Kit according to the manufacturer’ s instructions (https://verogen.com/documentation/). Briefly, library preparation included an initial two‐step PCR, using 1 ng of template DNA, to amplify the target loci and facilitate indexed adapter enrichment which help during the purification, and normalization of these libraries in the next step. Primer Mix B was used to amplify 58 STRs and 172 SNPs (not reported in this study). The prepared libraries were then pooled and denatured, and sequencing was performed using the Miseq FGx instrument (Verogen). Pooled libraries were placed in a Miseq FGx reagent cartridge and a flow cell facilitated the release of the incorporation buffers and sequencing reagents in accordance with the standard protocol. Five sequencing runs were performed and a negative and positive amplification control (2800 M, Verogen) was added to each run.

Sequence data analysis

Sequence data were analyzed using ForenSeqTM Universal Analysis Software (UAS) version 1.3 with Verogen's default settings. The analytical threshold (AT) and interpretation threshold (IT) were set at 1.5% and 4.5%, respectively. The STR intra‐locus balance threshold was set at 60% and the stutter filter was adjusted to reflect the specific needs of each locus. The minimum AT and IT were set at 10 and 30 reads, respectively, and the UAS used 650 reads as the minimum threshold before applying the AT and IT values. All sequence data were exported to Microsoft Office Excel and reviewed manually. All alleles identified in the sequencing analysis were then integrated into the Excel documents for further analysis.

Concordance study

Four commercial CE‐based STR kits were used to evaluate the overlap between STR genotypes generated using MPS and CE methods. Autosomal STRs (A‐STRs) were typed using a PowerPlex 21 System (Promega) and AGCU 21 + 1 Multiplex PCR Amplification Kits (AGCU), whereas Y chromosome STRs (Y‐STRs) and X chromosome STRs (X‐STRs) were typed using the Yfiler™ Platinum PCR Amplification (Thermo Fisher) and AGCU X19 Multiplex PCR Amplification Kits (AGCU), respectively. All 58 STRs (27 A‐STRs, 22 Y‐STRs, and 7 X‐STRs) were covered apart from DYS505 and DYS612. PCR products were separated and evaluated using an ABI 3500XL Genetic Analyzer (Thermo Fisher) according to the manufacturer's instructions. The electrophoretic results were analyzed using GeneMapper® ID‐X software v1.4 (Thermo Fisher). Any discordance within the CE‐based typing was evaluated using the binary sequence alignment (BAM) file and the Integrative Genomics Viewer (IGV) (Thorvaldsdóttir et al., 2013).

Identification of sequence variants

Sequence variation and allele frequencies were calculated and integrated into Excel according to the ISFG recommendations (Parson et al., 2016; Phillips, Devesse, et al., 2018; Phillips, Gettings, et al., 2018), to allow the comparison of these data with the records of the STR Sequencing Project (Gettings et al., 2017) and various other previous studies (Barrio et al., 2019; Churchill et al., 2017; Delest et al., 2020; Devesse et al., 2018, 2020; Gettings et al., 2015, 2016; Hussing et al., 2019; Jäger et al., 2017; Khubrani et al., 2019; Kim et al., 2016, 2018; Novroski et al., 2016; Peng et al., 2020; Phillips, Devesse, et al., 2018; Phillips, Gettings, et al., 2018; Wang et al., 2020; Wendt et al., 2017). Sanger sequencing was used to verify any novel alleles using a BigDye1 Terminator v3.1 Cycle Sequencing Kit (Thermo Fisher).

Forensic parameters and Y‐haplotype frequencies

The forensic parameters for all the A‐STRs were calculated using STRAF (Gouy & Zieger, 2017) and included by length (LB), sequence without flanking region (RSB), and sequence with flanking region (FSB). The forensic parameters included genotype count (N), allele count based on sequence (Nall), expected heterozygosity (Hexp) or genetic diversity (GD), polymorphism information content (PIC), random match probability (RMP), power of discrimination (PD), observed heterozygosity (Hobs), power of exclusion (PE), and typical paternity index (TPI). Allele frequencies (LB, RSB, and FSB,) were calculated in Excel (allele count/total). STRAF was also used to test for Hardy–Weinberg equilibrium (HWE), applying a Bonferroni correction for multiple comparisons.

The Y‐haplotype frequencies were calculated using the direct counting method incorporating LB, RSB, and FSB alleles. RMP was calculated using the equation P = ∑χ2, whereas discrimination power (DP) was calculated using DP = 1−∑χ2 and haplotype diversity (HD) was calculated using HD = n(1‐∑χ2)/n−1, where χ is the frequency of each Y‐STR haplotype.

RESULTS AND DISCUSSION

Sequencing results

This study produced 17,510,320 reads over five sequencing runs. No allele above the interpretation threshold was detected in any of the negative amplification controls. Sequencing metrics for each run are shown in Table S1. Data analyzed by UAS were exported to Microsoft Office Excel sheets for manual review.

Here, we identified a heterozygous imbalance in D22S1045 which may cause some allele dropout resulting in some heterozygotes appearing to be homozygotes. For this reason, no further analysis (concordance study, allele frequency, and forensic parameters) was performed for this locus. In addition, we identified a high rate of allelic dropout at DYS392 (5%, 6/120). The poor performance at these two loci is due to the data quality issues/interpretation challenges, regardless of total sample read counts according to previous reports (Just et al., 2017; Novroski et al., 2016; Peng et al., 2020), which was also noted by the manufacturer's protocol (Verogen, 2018). Besides D22S1045 and DYS392, the instances of allelic dropout were also observed at another three loci: PentaE (2/240), DYS448 (1/120), and DXS7132 (1/120). A complete list of all A‐STRs and Y‐STRs alleles identified in 120 individuals is detailed in Tables S3 and S5.

Concordance refers to the likelihood of obtaining the same allele calls using any one of multiple methods, in this case MPS and CE. Before being implemented in routine casework, MPS data should be shown to be compatible with current CE‐based forensic databases (Devesse et al., 2018, 2020). It is crucial to evaluate the consistency between these two methods. In our study, all loci in the ForenSeq DNA Signature Prep Kit were compared except for D22S1045, as explained above, and DYS505 and DYS612, which were not included in all of the CE kits used in this study. The CE‐based profiles of all A‐STRs and Y‐STRs are detailed in Tables S2 and S4.

Our data suggest that there is over 99% concordance between each of the data points evaluated in this experiment. In the instances of allelic dropout mentioned above, alleles were obtained by CE. We assumed that the allelic dropout at DYS392 was due to the defects of the ForenSeq DNA Signature Prep Kit. The allelic drop out at the other loci was likely associated with SNPs or other mutations within the MPS primer binding sites.

One discordance, at DYS439, was observed. With the exception of DYS385a‐b and DYS387S1, Y‐STRs are expected to present with only one allele. However, our sample data suggested that there was a duplicated allele at DYS439. This duplication was only detected by MPS, whereas the CE‐based genotyping appeared normal. This discordance could have been caused by differences in the primer sequences used for the MPS and CE assays, as described by Kwon et al. (2016).

There was an additional discordance at the PentD allele where the CE‐based genotyping called a 9.2, 11 pattern and the MPS‐based genotyping called a 10, 11 ([AAAGA]10, [AAAGA]11). The allele with the [AAAGA]10 repeat was subsequently confirmed after the BAM file was checked using the IGV program. No sequence variation in the flanking regions was detected using UAS. The sample was sequenced using primers designed to extend the detection in the flanking region. A rare indel (rs1176142838) in the downstream flanking region was observed, which was determined to cause the discordance. The explanation is that the position of the three bases (TAA) deletion is outside of the bioinformatic recognition sites but within the CE amplified region. The original sequence data showed no discordance in the length of the amplicons between MPS‐based and CE‐based genotype. Thus, the discordance was due to bioinformatics configurations, as described by Gettings et al. (2016). Overall, the concordance rate between the two methods is extremely high and we can assume that the accidental discordances can be further reduced by improving bioinformatics analysis methods and expanding the available pool of MPS‐STR population data.

Sequence variation and diversity of observed alleles

Without considering D22S1045, the instances of allelic dropout, duplicated Y‐STR alleles, and the discordance described above, we were able to identify 10187 alleles from 120 individuals (A‐STRs: 6236 alleles, Y‐STRs: 3112 alleles, and X‐STRs: 839 alleles). MPS analysis of these STRs substantially increased the allele diversity, as demonstrated in previous studies (Barrio et al., 2019; Churchill et al., 2017; Delest et al., 2020; Gettings et al., 2015, 2016; Hussing et al., 2019; Jäger et al., 2017; Khubrani et al., 2019; Kim et al., 2016, 2018; Novroski et al., 2016; Peng et al., 2020; Phillips, Devesse, et al., 2018; Phillips, Gettings, et al., 2018; Wang et al., 2020; Wendt et al., 2017). The increase in the number of unique alleles varied from locus to locus.

For the A‐STR loci, 235 unique LB alleles, 325 unique RSB alleles, and 363 unique FSB alleles were observed. With sequence variation in the repeat and flanking regions identified in over 69% of the loci (18/26) demonstrating an increase in the number of unique alleles identified using MPS versus CE. D21S11 exhibited the highest diversity, with 33 unique alleles and TPOX displaying the lowest diversity, with only five unique alleles. The number of unique alleles per locus by LB, RSB, and FSB was compared to each other, as shown in Table 1a and Figure 1. Eight loci showed an increase in allele diversity due to variation within the repeat region sequence alone (D1S1656, D2S1338, D3S1358, D4S2408, D8S1179, D9S1122, D12S391, and FGA), whereas three loci exhibited increased allele diversity due to variations in only the flanking region (D5S818, D10S1248, and D16S539) and seven loci displayed an increase in allele diversity due to variations in the repeat and flanking region sequences (D2S441, D6S1043, D7S820, D13S317, D20S482, D21S11, and vWA). The eight remaining loci demonstrated no increase in allele diversity (D17S1301, D18S51, D19S433, CSF1PO, PentaD, PentaE, TH01, and TPOX). These results are similar but slightly different from those of previous studies (Barrio et al., 2019; Churchill et al., 2017; Delest et al., 2020; Gettings et al., 2015, 2016; Hussing et al., 2019; Jäger et al., 2017; Khubrani et al., 2019; Kim et al., 2016, 2018; Novroski et al., 2016; Peng et al., 2020; Phillips, Devesse, et al., 2018; Phillips, Gettings, et al., 2018; Wang et al., 2020; Wendt et al., 2017). Gettings et al. (2016) reported an increase in allele diversity at D19S433 and PentaE, whereas no sequence variation was observed at D7S820, D13S317, and D16S539 in their study (three populations, N = 183). Sequence variation was observed at D19S433 and PentaD in the study by Delest et al. (2020) (French population, N = 169), whereas in the study by Khubrani et al. (2019), sequence variation was observed at TH01 but not at D6S1043 (Arab population, N = 89). Novroski et al. (2016) reported sequence variations in all the A‐STRs except TPOX (four populations, N = 777). This suggests that sequence variations differ slightly between different populations, and these slight differences are likely due to the differences in genetic admix and the number of samples investigated.

TABLE 1A

The number of unique alleles observed at 26 A‐STRs by LB, by RSB, and by FSB

Locus	LB	RSB	FSB	Increase	%Increase
D12S391	11	30	30	19	172.73
D21S11	13	32	33	20	153.85
D13S317	8	9	20	12	150.00
D2S1338	11	25	25	14	127.27
D7S820	8	10	18	10	125.00
D3S1358	6	13	13	7	116.67
D20S482	6	8	13	7	116.67
D8S1179	9	17	17	8	88.89
D16S539	7	7	13	6	85.71
D2S441	7	10	12	5	71.42
D1S1656	9	14	14	5	55.56
D9S1122	8	12	12	4	50.00
D5S818	7	7	10	3	42.86
D10S1248	6	6	8	2	33.33
vWA	8	10	10	2	25.00
D4S2408	5	6	6	1	20.00
D6S1043	15	17	17	2	13.33
FGA	15	16	16	1	6.67
D17S1301	6	6	6	0	0.00
D18S51	13	13	13	0	0.00
D19S433	11	11	11	0	0.00
CSF1PO	8	8	8	0	0.00
PentaD	9	9	9	0	0.00
PentaE	18	18	18	0	0.00
TH01	6	6	6	0	0.00
TPOX	5	5	5	0	0.00
Total	235	325	363	128	54.47

Abbreviations: A‐STRs, Autosomal STRs; FSB, sequence‐based alleles with flanking region; LB, length‐based alleles; RSB, sequence‐based alleles without flanking region.

FIGURE 1

Allele diversity of A‐STRs based on length or sequence (with or without flanking regions)

There were 145 unique LB alleles, 203 unique RSB alleles, and 212 unique FSB alleles identified in our Y‐STR data. While there were 61 unique LB alleles, 78 unique RSB alleles, and 78 unique FSB alleles within the X‐STR samples. The increases in diversity for both the Y‐ and X‐STRs were smaller than that of the A‐STR loci. Approximately 53.8% of Y‐STRs (14/26) presented with increased allelic diversity, whereas only 42.9% (3/7) of the X‐STR loci exhibited any increases in diversity. Among these loci, nine of the Y‐STR loci gained in allele diversity due to variation in the repeat region sequence alone (DYF387S1, DYS389I, DYS389II, DYS439, DYS448, DYS481, DYS612, and DYS635). Three exhibited increased diversity caused by changes in the flanking regions alone (DYS19, DYS460, and Y‐GATA‐H4) and two Y‐STR loci displayed an increase in allele diversity due to variations in both the repeat and flanking region sequences (DYS390 and DYS437). All X‐STR loci with increased allele diversity displayed an increase in variation within the repeat region sequence (DXS10135, DXS7132, and DXS10103) but no other changes were evident. DYF387S1 presented with the highest diversity among the Y‐STR loci, with 26 unique alleles and DXS10135 was the most variable X‐STR locus, with 34 unique alleles. The lowest Y‐STR and X‐STR diversity was observed at DYS391, with three unique alleles, and DXS7423, with five unique alleles, respectively. The number of alleles per locus for LB, RSB, and FSB is summarized in Tables 1b,c, and Figure 2.

TABLE 1B

The number of unique alleles observed at 24 Y‐STRs and 7 X‐STRs by LB, by RSB, and by FSB.

Locus	LB	RSB	FSB	Increase	%Increase
DYS389II	7	21	21	14	200.00
DYF387S1	9	26	26	17	188.89
DYS390	7	14	16	9	128.57
DYS437	4	7	8	4	100.00
DYS448	6	12	12	6	100.00
Y‐GATA‐H4	4	4	8	4	100.00
DYS635	6	10	10	4	66.67
DYS612	9	12	12	3	33.33
DYS460	4	4	5	1	25.00
DYS19	5	5	6	1	20.00
DYS389I	5	6	6	1	20.00
DYS439	5	6	6	1	20.00
DYS481	13	15	15	2	15.38
DYS385a‐b	12	12	12	0	0.00
DYS391	3	3	3	0	0.00
DYS392	6	6	6	0	0.00
DYS438	4	4	4	0	0.00
DYS505	4	4	4	0	0.00
DYS522	5	5	5	0	0.00
DYS533	5	5	5	0	0.00
DYS549	4	4	4	0	0.00
DYS570	7	7	7	0	0.00
DYS576	5	5	5	0	0.00
DYS643	6	6	6	0	0.00
Total	145	203	212	67	46.215

Abbreviations: FSB, sequence‐based alleles with flanking region; LB, length‐based alleles; RSB, sequence‐based alleles without flanking region; Y‐STRs, Y chromosome STRs.

TABLE 1C

The number of unique alleles observed at 7 X‐STRs by LB, by RSB, and by FSB.

Locus	LB	RSB	FSB	Increase	%Increase
DXS10135	20	34	34	14	70.00
DXS7132	8	10	10	2	25.00
DXS10103	6	7	7	1	16.67
DXS8378	6	6	6	0	0.00
DXS10074	9	9	9	0	0.00
DXS7423	5	5	5	0	0.00
HPRTB	7	7	7	0	0.00
Total	61	78	78	17	27.87

Abbreviations: FSB, sequence‐based alleles with flanking region; LB, length‐based alleles; RSB, sequence‐based alleles without flanking region; X‐STRs, X chromosome STRs.

FIGURE 2

Allele diversity of Y‐ and X‐STRs based on length or sequence (with or without flanking regions)

Allele frequencies and forensic parameters

The observed LB, RSB, and FSB allele frequencies for all the STR loci (except D22S1045) from 120 Tibetan individuals were calculated using the counting method, as summarized in Tables S6–S8 (with those instances of discordance described above not included). All A‐STR loci allele data met HWE expectations after Bonferroni correction (α=0.05/26), as listed in Table S9, which means that the population data from this study can be considered representative.

The forensic parameters (described in Material and Methods) for each A‐STR locus obtained from the LB, RSB, and FSB data were evaluated using STRAF and are summarized in Table S9. The average GD/Hexp for all of the A‐STRs was 0.7785 when analyzed by length, 0.7979 when evaluated using the repeat region sequence, and 0.8106 when sequence variation in the flanking region was considered. The combined RMP for the 26 A‐STRs was 2.04 × 10–29, 1.93 × 10–31, and 9.56 × 10–33 for LB, RSB, and FSB, respectively. When we used the sequence variation in the repeat regions alone, the combined RMP was more than 105 times lower than that of the length‐based alleles. The addition of sequence variation in the repeat and flanking regions, allowed MPS analysis to reduce the combined RMP by over 2100 times compared to the CE method. Similar trends were observed in other forensic parameters resulting from the increase in the number of unique alleles available. A total of 111 and 113 unique haplotypes in the Y‐STR loci were observed when using length‐based and sequence‐based alleles, respectively. Sequence variations in the flanking region did not increase the number of unique haplotypes. While among the 111 unique length‐based allele haplotypes, 106, 5, and 2 were observed once, twice, and three times, respectively (Table S4). The RMP was 0.01, the DP was 0.99, and the HD was 0.9983. When sequence variation was considered, 108, 3, and 2 haplotypes were observed once, twice, and three times, respectively (Table S5) whereas RMP was reduced to 0.0096, DP was increased to 0.9904, and HD increased to 0.9987. Overall, the advantages of the MPS are reflected in the forensic parameters.

Novel alleles and SNPs in flanking regions

Among the 653 unique sequence‐based alleles (including the flanking region) observed in this study, 35 identified in 25 loci (12 A‐STRs loci, 11 Y‐STR loci, and 2 X‐STRs loci, respectively) were not recorded in the STR Sequencing Project (Gettings et al., 2017) and have never been reported in any of the previous studies (Barrio et al., 2019; Churchill et al., 2017; Delest et al., 2020; Devesse et al., 2018, 2020; Gettings et al., 2015, 2016; Hussing et al., 2019; Jäger et al., 2017; Khubrani et al., 2019; Kim et al., 2016, 2018; Novroski et al., 2016; Peng et al., 2020; Phillips, Devesse, et al., 2018; Phillips, Gettings, et al., 2018; Wang et al., 2020; Wendt et al., 2017). Sanger sequencing was performed to verify these novel sequence‐based alleles (data not provided) and all 35 are listed in Table 2.

TABLE 2

Thirty‐five novel alleles of STR loci observed in this study.

Locus	MPS allele (following the nomenclature recommended by IFSG)
D5S818	D5S818 [CE6]‐GRCh38‐Chr5‐123775543‐123775606 [ATCT]6
D6S1043	D6S1043 [CE20.3]‐GRCh38‐Chr6‐91740160‐91740292 [ATCT]6[ATGT][ATCT]2[ATC][ATCT]11 91740273‐A
D6S1043	D6S1043 [CE21.3]‐GRCh38‐Chr6‐91740160‐91740292 [ATCT]6[ATGT][ATCT]2[ATC][ATCT]12 91740273‐A
D7S820	D7S820 [CE11]‐GRCh38‐Chr7‐84160191‐84160297 [TATC]9[TGTC][TATC] 84160204‐A
D9S1122	D9S1122 [CE17]‐GRCh38‐Chr9‐77073809‐77073880 [TAGA][TCGA][TAGA]15
D10S1248	D10S1248 [CE14]‐GRCh38‐Chr10 129294226‐129294318 [GGAA]14 129294238‐A
D10S1248	D10S1248 [CE15]‐GRCh38‐Chr10 129294226‐129294318 [GGAA]15 129294243‐A
D12S391	D12S391 [CE21]‐GRCh38‐Chr12‐12296981‐12297189 [AGAT]4[AGGT][AGAT]9[AGAC]6[AGAT]
D13S317	D13S317 [CE14]‐GRCh38‐Chr13‐82147986‐82148107 [TATC]14 82148069‐T 82148073‐T
D13S317	D13S317 [CE15]‐GRCh38‐Chr13‐82147986‐82148107 [TATC]15 82148069‐T
D16S539	D16S539 [CE14]‐GRCh38‐Chr16‐86352664‐86352781 [GATA]14 86352761‐C
D18S51	D18S51 [CE7]‐GRCh38‐Chr18‐63281662‐63281796 [AGAA]7
D20S482	D20S482 [CE14]‐GRCh38‐Chr20‐4525674‐4525771 [AGAT]4[ATAT][AGAT]9
D20S482	D20S482 [CE15]‐GRCh38‐Chr20‐4525674‐4525771 [AGAT]14[AGAC]
D21S11	D21S11 [CE29]‐GRCh38‐Chr21‐19181939‐19182111 [TCTA]6[TCTG]5[TCTA]3 TA [TCTA]3 TCA[TCTA]2 TCCATA [TCTA]10 19182101‐T
D21S11	D21S11 [CE30.2]‐GRCh38‐Chr21‐19181939‐19182111 [TCTA]5[TCTG]7[TCTA]3 TA [TCTA]2 TCA[TCTA]2 TCCATA [TCTA]10 TA[TCTA]
D21S11	D21S11 [CE31.2]‐GRCh38‐Chr21‐19181939‐19182111 [TCTA]5[TCTG]7[TCTA]2 TA [TCTA]3 TCA[TCTA]2 TCCATA [TCTA]11 TA[TCTA]
D21S11	D21S11 [CE33.2]‐GRCh38‐Chr21‐19181939‐19182111 [TCTA]5[TCTG]6[TCTA]3 TA [TCTA]4 TCA[TCTA]2 TCCATA [TCTA]12 TA[TCTA]
FGA	FGA [CE29]‐GRCh38‐Chr4‐154587713‐154587840 [GGAA]2[GGAG][AAAG]21[AGAA][AAAA][GAAA]3
DYS19	DYS19 [CE14]‐GRCh38‐ChrY‐9684267‐9684443 [TCTA]11 CCTA [TCTA]3 9684269‐T
DYS390	DYS390 [CE19]‐ChrY‐GRCh38 15162096‐15163170 [TAGA]11[CAGA]8
DYS390	DYS390 [CE23]‐ChrY‐GRCh38 15162096‐15163170 [TAGA]3[CAGA][TAGA]10[CAGA]9
DYS389I	DYS389I [CE13]‐ChrY‐GRCh38 12500387‐12500513 [TAGA]9[CAGA]4
DYS389II	DYS389II [CE29]‐ChrY‐GRCh38 12500448‐12500633 [TAGA]9[CAGA]4N48 [TAGA]11[CAGA]5
DYS439	DYS439 [CE11]‐ChrY‐GRCh38 12403461‐12403587 [GATA]11 12403513‐G 12403514‐A 12403515‐T
DYS460	DYS460 [CE9]‐ChrY‐GRCh38 18888810‐18889046 [TATC]9 1888811‐G
DYS481	DYS481 [CE17]‐ChrY‐GRCh38 8558313‐8558408 [CTT]17
DYS635	DYS635 [CE23]‐ChrY‐GRCh38 12258755‐12258975 [TAGA]12 [TACA]3 [TAGA]2 [TACA]2 [TAGA]4
DYF387S1	DYF387S1 [CE40]‐GRCh38‐ChrY‐23785347‐23785521 [AAAG]3[GTAG][GAAG]4[AAAG]2[GAAG][AAAG]2[GAAG]8[AAAG]19
DYF387S1	DYF387S1 [CE42]‐GRCh38‐ChrY‐23785347‐23785521 [AAAG]3[GTAG][GAAG]4[AAAG]2[GAAG][AAAG]2[GAAG]10[AAAG]19
Y‐GATA‐H4	Y‐GATA‐H4 [CE10]‐ChrY‐GRCh38 16631624‐16631759 [TCTA]10 16631721‐T
DXS10135	DXS10135 [CE14]‐GRCh38‐ChrX‐9338302‐9338520 [AAGA]3 N7 [AAGA]10[AAAG]
DXS10135	DXS10135 [CE26]‐GRCh38‐ChrX‐9338302‐9338520 [AAGA]3 N7 [AAGA]15[AAGG][AAGA]4[AAGG][AAGA][AAAG]
DXS10135	DXS10135 [CE30]‐GRCh38‐ChrX‐9338302‐9338520 [AAGA]3 N7 [AAGA]19[AAGG]2[AAGA]3[AAGG][AAGA][AAAG]
DXS7132	DXS7132 [CE16]‐GRCh38‐ChrX‐65435623‐65435778 [TAGA]15[CAGA]

Abbreviations: CE, capillary electrophoresis; MPS, massively parallel sequencing.

In total, 24 SNPs and one InDel were observed in the flanking region of 17 STRs (Table 3) 6 of which were not recorded in the Single‐Nucleotide Polymorphism database (dbSNP, https://www.ncbi.nlm.nih.gov/snp/). Among the 25 sequence variations in the flanking region, 15 SNPs were found within 10 A‐STRs, whereas variations in the flanking regions of the Y‐STRs and X‐STRs were not as frequent. Nine SNPs and one deletion were observed in the flanking regions of six Y‐STR loci and one X‐STR locus, respectively.

TABLE 3

SNPs and InDels observed in flanking regions at 58 STRs using UAS.

Locus	Variation	Position (GRCh38/hg38)	dbSNP ID	Wild	Mutant	Count
D2S441	SNP	Chr2: 68011922	rs74640515	G	A	25
D5S818	SNP	Chr5: 123775552	rs73801920	C	A	54
D6S1043	SNP	Chr6: 91740273	rs529713981	G	A	4
D7S820	SNP	Chr7: 84160204	rs7789995	T	A	222
D7S820	SNP	Chr7: 84160286	rs16887642	G	A	42
D10S1248	SNP	Chr10: 129294238	rs1279061683	G	A	1
D10S1248	SNP	Chr10: 129294243	rs563636310	T	A	1
D13S317	SNP	Chr13: 82148069	rs9546005	A	T	120
D13S317	SNP	Chr13: 82148073	rs202043589	A	T	23
D16S539	SNP	Chr16: 86352692	rs563997442	C	G	2
D16S539	SNP	Chr16: 86352761	rs11642858	A	C	77
D20S482	SNP	Chr20: 4525681	rs561985213	G	A	2
D20S482	SNP	Chr20: 4525680	rs77560248	C	T	29
D21S11	SNP	Chr21: 19182101	rs1051967683	C	T	1
vWA	SNP	Chr12: 5983970	rs75219269	A	G	1
DYS19	SNP	ChrY: 9684269	Null	G	T	1
DYS390	SNP	ChrY: 15163163	rs758940870	T	C	2
DYS437	SNP	ChrY: 12346421	Null	G	A	9
DYS439	SNP	ChrY: 12403513	rs1042036966	A	G	1
DYS439	SNP	ChrY: 12403514	Null	G	A	1
DYS439	SNP	ChrY: 12403515	Null	A	T	1
DYS460	SNP	ChrY: 1888811	Null	T	G	1
Y‐GATA‐H4	SNP	ChrY: 16631721	rs765275581	C	T	1
Y‐GATA‐H4	SNP	ChrY: 16631756	Null	A	G	59
DXS10135	Deletion	ChrX: 9338410–9338416	rs201630737	AAGAAGA	AGA	1

Abbreviations: dbSNP, Single‐Nucleotide Polymorphism database; Null, No record in dbSNP.

Most sequence variations in the flanking regions were found to be present in low proportions. However, SNPs in the flanking regions of some of the A‐STR loci were so frequent that they could be observed in a large proportion of sequence‐based alleles. D7S820 was shown to exhibit the highest proportion of sequence‐based alleles with flanking SNPs (rs7789995: 222/240 and rs16887642: 42/240) with D13S317 following a similar pattern (rs9546005: 120/240 and rs202043589: 23/240). The variations at Y‐GATA‐H4 (ChrY: 16631756) were observed in almost half of the samples (59/120), although this SNP was not recorded in the dbSNP. DYS439 also had an interesting SNP allele, DYS439 [CE11]‐ChrY‐GRCh38 12403461‐12403587 [GATA]11 12403513‐G 12403514‐A 12403515‐T) which included three coterminous SNPs within its flanking region. This allele can also be referred to as DYS439 [CE11]‐ChrY‐GRCh38 12403461‐12403587 [GATA]12 12403513‐12403516 Del, which suggests that there was an additional repeat [GATA] within the repeat region and the sequence variation in the flanking region was a four base (AGAA) deletion. Here, we use the first name to allow overlap with the CE databases. These novel alleles and polymorphisms in the flanking regions can further enrich the MPS‐STR population data.

CONCLUSION

Here, we report the MPS‐STR profile data for the Tibetan population. The diversity of the observed alleles, allele frequencies, and forensic parameters per locus by length, sequence without flanking region, and sequence with flanking region were analyzed and compared and clearly demonstrate the advantages of MPS in comparison to CE. To ensure compatibility between the MPS data and the current CE‐based forensic databases, we completed a concordance study that demonstrated a concordance rate of more than 99% with some exceptions which were subjected to further analysis. Our data suggest that MPS‐ and CE‐derived alleles are compatible and we were able to identify 35 novel alleles at 25 loci using the MPS method.

In conclusion, with more pooled data and larger‐scale validation, MPS could play a valuable role in forensic genetics and might be an additional tool for routine casework.

CONFLICT OF INTEREST

The authors have no conflicts of interest to declare.

AUTHORS’ CONTRIBUTIONS

J.X. and X.Z. involved in the study design. C.Z. involved in sample collection. K.M., Y.C., and Q.Y. involved in experimental work. H.L. and G.S. involved in data analysis. H.L. involved in the writing of the initial manuscript. J.X. and X.Z. involved in revision. All authors reviewed the manuscript.

Tables S1‐S9

Click here for additional data file.