Advertisement

Evaluation of sensitivity and specificity of sibship determination in the Caucasian population of the Russian Federation using the 23 STR loci VeriFiler panel

Published:September 23, 2019DOI:https://doi.org/10.1016/j.fsigss.2019.09.023

      Abstract

      We performed a simulation study to evaluate sensitivity and specificity of sibship determination in the Caucasian population of the Russian Federation using 23 autosomal STR loci included in the VeriFiler Express (Applied Biosystems) kit. Population genetic data were obtained from 1609 unrelated Caucasian individuals from different regions of the Russian Federation. Several scenarios were analyzed: half siblings duo vs unrelated; half siblings vs unrelated (mother(s) will be genotyped); full siblings duo vs half siblings; full siblings vs half siblings (mother will be genotyped); full siblings duo vs unrelated.

      Keywords

      1. Introduction

      The 13th edition of Standards for Relationship Testing Laboratories [
      Standards for Relationship Testing Laboratories.
      ] contains a new requirement to report “the estimate of the percentage of individuals of known relationship that may have a combined likelihood ratio that is inconclusive, or supportive, or not supportive of the tested relationship” for two-party comparisons of full sibling, half sibling, avuncular, and single grandparent.
      We performed a simulation study to evaluate sensitivity and specificity of sibship determination in the Caucasian population of the Russian Federation using 23 autosomal STR loci included in the VeriFiler Express (Applied Biosystems) kit (D3S1358, vWA, D16S539, CSF1PO, TPOX, D8S1179, D21S11, D18S51, D2S441, D19S433, TH01, FGA, D22S1045, D5S818, D13S317, D7S820, D10S1248, D1S1656, D12S391, D2S1338, D6S1043, Penta D, Penta E).

      2. Materials and methods

      Population genetic data were obtained from 1609 unrelated Caucasian individuals from different regions of the Russian Federation (admixed urban population). Allelic frequencies, forensic and population genetic parameters were determined using the STRAF v1.0.5 software package [
      • Gouy A.
      • Zieger M.
      STRAF - A convenient online tool for STR data evaluation in forensic genetics.
      ]. For F-statistics, 8 regional subpopulations were defined: Altai Area (198 samples), Amur Region (116 samples), Belgorod Region (130 samples), Bryansk Region (276 samples), Vladimir Region (114 samples), Volgograd Region (424 samples), Ivanovo Region (257 samples), and Kursk Region (94 samples). All selected subgroups included at least 90 individuals each.
      The simulation study was performed using the Familias v3.2.7 software package [
      • Kling D.
      • Tillmar A.O.
      • Egeland T.
      Familias 3 - Extensions and new functionality.
      ]. Different scenarios were analyzed: half siblings duo vs unrelated; half siblings vs unrelated (mother(s) will be genotyped); full siblings duo vs unrelated; full siblings duo vs half siblings; full siblings vs half siblings (mother will be genotyped). Genotypic configurations were simulated assuming silent allele frequency 0.005 at each locus. Mutations were not considered. Percentage of true and false positives was determined using the likelihood ratio (LR) threshold of 10, percentage of true and false negatives was determined using the LR threshold of 0.1 [
      Standards for Relationship Testing Laboratories.
      ].

      3. Results and discussion

      Population genetic parameters of our dataset relevant to forensics and kinship testing are given in Table 1.
      Table 1Population genetic data and forensic parameters of the 23 STR loci Verifiler panel in the Caucasian population of the Russian Federation (n = 1609). Nall – number of different alleles observed; GD – gene diversity (expected heterozygosity); PIC – polymorphism information content, PM – match probability; PD – power of discrimination; Hobs - observed heterozygosity; PE – power of exclusion; TPI – typical paternity index; FST – subpopulation correction; FIS – coefficient of inbreeding; pHW – p-values for deviation from the Hardy-Weinberg equilibrium.
      LocusNallGDPICPMPDHobsPETPIFSTFISpHW
      CSF1PO100.7410.6950.1140.8860.7380.4891.9061.7E-050.01590.216
      D10S124890.7630.7240.0950.9050.7560.5202.047−1.2E-030.02330.157
      D12S391160.8860.8750.0250.9750.8800.7554.168−7.9E-040.00130.121
      D13S317100.7800.7510.0780.9220.7910.5822.387−5.0E-05−0.02410.663
      D16S53980.7650.7290.0950.9050.7750.5542.222−5.7E-04−0.00610.493
      D18S51180.8770.8650.0280.9720.8810.7564.190−3.7E-04−0.0120.100
      D19S433160.7850.7580.0750.9250.7890.5782.3664.7E-04−0.01930.595
      D1S1656210.9010.8920.0190.9810.8970.7894.846−4.6E-040.00540.150
      D21S11180.8520.8350.0390.9610.8570.7093.4981.1E-04−0.00910.197
      D22S1045100.7430.7010.1110.8890.7480.5061.9822.7E-04−0.00880.913
      D2S1338150.8820.8700.0260.9740.8810.7564.1906.0E-04−0.0020.568
      D2S441120.7480.7070.1040.8960.7440.4991.953−2.0E-040.02070.794
      D3S1358100.7860.7510.0800.9200.7810.5642.2793.8E-04−0.00140.661
      D5S818100.7310.6870.1170.8830.7270.4701.828−6.8E-040.00130.909
      D6S1043150.8300.8110.0490.9510.8140.6262.6914.8E-040.02020.266
      D7S820100.8010.7730.0680.9320.7870.5762.3523.2E-040.01120.292
      D8S1179110.8030.7760.0660.9340.8070.6122.587−3.5E-041.00E-040.249
      FGA220.8620.8470.0350.9650.8650.7243.690−4.2E-068.00E-040.174
      Penta D180.8260.8020.0550.9450.8360.6673.047−7.7E-04−0.0050.910
      Penta E190.9030.8940.0180.9820.8920.7804.6503.9E-040.00690.201
      TH0190.7780.7420.0850.9150.7770.5572.241−3.8E-04−0.00280.922
      TPOX80.6050.5550.2080.7920.6050.2971.2673.2E-050.00670.512
      vWA90.8070.7790.0660.9340.8100.6172.6291.9E-04−0.0070.344
      All loci are in the Hardy-Weinberg equilibrium.
      Table 2 presents results of simulations for each scenario tested (100,000 simulations for each combination of the true hypothesis and the number of testing participants).
      Table 2Simulation results. HS – half sibling, FS – full sibling.
      ScenarioTrue hypothesisLRPercentage of simulations with:
      Median95%5%Standard deviationLR > 10 (supporting H1), %0.1 < LR < 10 (inconclusive), %LR < 0.1 (supporting H2), %
      HS (H1) vs. Unrelated (H2)H168.291.528e+0040.50152.595e+00673.57625.0691.355
      H20.020951.7220.000335126.781.20926.44972.342
      HS (H1) vs. Unrelated (H2)H1751.57.263e+0051.3693.5e+00986.78912.3390.872
      H20.0034110.64273.48e-00511650.71613.30985.975
      Data one mother
      HS (H1) vs. Unrelated (H2)H11.8e+0049.585e+0076.5917.623e+01194.0215.4610.518
      H20.00049140.1634.167e-00683.170.3616.11893.521
      Data both mothers
      FS (H1) vs. HS (H2)H1132.25.891e+0040.52566.199e+00677.12121.4351.444
      H20.018991.3980.000572532.11.01524.10174.884
      FS (H1) vs. HS (H2)H12.455e+0041.432e+0087.8617.567e+01194.4565.0710.473
      H20.00039560.14063.203e-00633.650.3435.58294.075
      Data mother
      FS (H1) vs. Unrelated (H2)H16.11e+0062.832e+011288.25.936e+01798.6271.2750.098
      H24.06e-0060.006744.675e-00978.80.0721.29398.635

      4. Conclusions

      For half siblings duo, simulating the true relationship, 73.576% of the 100,000 simulations were equal or above the LR limit, 1.355% of false negatives; simulating the alternative hypothesis yielded 72.342% of true negatives and 1.209% of false positives. For half siblings (one mother will be genotyped), simulating the true relationship, 86.789% of the 100,000 simulations were equal or above the LR limit, 0.872% of false negatives; simulating the alternative hypothesis yielded 85.975% of true negatives and 0.716% of false positives. For half siblings (mothers will be genotyped), simulating the true relationship, 94.021% of the 100,000 simulations were equal or above the LR limit, 0.518% of false negatives; simulating the alternative hypothesis yielded 93.521% of true negatives and 0.361% of false positives. For full siblings duo vs. half siblings, simulating the true relationship, 77.121% of the 100,000 simulations were equal or above the LR limit, 1.444% of false negatives; simulating the alternative hypothesis yielded 74.884% of true negatives and 1.015% of false positives. For full siblings vs. half siblings (mother will be genotyped), simulating the true relationship, 94.456% of the 100.000 simulations were equal or above the LR limit, 0.473% of false negatives; simulating the alternative hypothesis yielded 94.075% of true negatives and 0.343% of false positives. For full siblings vs. unrelated, simulating the true relationship, 98.627% of the 100,000 simulations were equal or above the LR limit, 0.098% of false negatives; simulating the alternative hypothesis yielded 98.635% of true negatives and 0.072% of false positives.

      Declaration of Competing Interest

      None.

      Acknowledgements

      The authors would like to thank G. Kostinyuk, I. Kalambet, and N. Kalambet for laboratory assistance with DNA analysis.

      References

      1. Standards for Relationship Testing Laboratories.
        13th ed. AABB, 2017
        • Gouy A.
        • Zieger M.
        STRAF - A convenient online tool for STR data evaluation in forensic genetics.
        Forensic Sci. Int. Genet. 2017; 30: 148-151
        • Kling D.
        • Tillmar A.O.
        • Egeland T.
        Familias 3 - Extensions and new functionality.
        Forensic Sci. Int. Genet. 2014; 13: 121-127