If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Corrigendum to “Evaluation of the IrisPlex DNA-based eye color prediction assay in a United States population” [Forensic Sci. Int. Genet. (2014) 111–117]
Department of Biology and Forensic and Investigative Sciences Program, Indiana University-Purdue University Indianapolis, 723 W. Michigan Street, Indianapolis, IN 46202, USA
Department of Biology and Forensic and Investigative Sciences Program, Indiana University-Purdue University Indianapolis, 723 W. Michigan Street, Indianapolis, IN 46202, USA
After further scrutiny of the genotype and eye color data of the 200 samples used in the dataset while analyzing it for post-published purposes, some inconsistencies were noted which the authors would like to further report here. There were 9 samples (5 in training set, 4 in the validation set) that had their genotypes shifted by one sample in the data so that the true genotype of the sample was not correctly aligned. The erroneously shifted genotypes were corrected and allele frequencies were recalculated for 4 SNPs. These corrections were made to the model parameters and all analyses repeated.
Maximum prediction accuracies following allele frequency adjustment went from 58% to 77% for brown, from 95% to 79% for blue, and 11% to 65% for intermediate (Table 3). The figures and tables were corrected and are shown below (Fig. 2, Fig. 3 and Table 3, Table 4, Table 5). Briefly to summarize the changes, there was a small increase of sensitivity for blue eye color with all models with an increase from 93% to 95%, and a decrease for brown and intermediate eye colors, 97% to 89% and 93% to 0%, respectively (Table 5). For all models, there was an increase in correct prediction rates for intermediate eye color from 11% to 65% and blue eye color from 55% to 79%, but a decrease in brown eye color from 98% to 57%. Overall frequencies of predictions (Fig. 2, Fig. 3) mostly decreased in inconclusive results, but also increased in the number of incorrect predictions. AUC values were approximately the same across all models, except with a decrease in intermediate from 0.88 to 0.77 (Table 4). There was no overall change from original conclusions made that the Bayesian network model should still be considered as an optimal prediction model method (over MLR), this is especially true for intermediate sample predictions.
Table 3The corrected prediction rates (%) by color category of the verification set (N = 100) was evaluated against the IrisPlex regression parameters and the adjusted regression parameters. The verification set was then evaluated using the Bayesian network with either set of a priori odds.
Parameters
Threshold
Original
Corrected
Brown (%)
Intermediate (%)
Blue (%)
Brown (%)
Intermediate (%)
Blue (%)
MLR: IrisPlex
0.5
88
0
95
90
0
93
0.7
76
0
95
79
0
91
MLR: Adjusted
0.5
58
19
93
79
53
77
0.7
42
11
95
79
41
64
Bayesian: Equal oddsc
0.5
55
20
80
79
65
64
0.7
55
20
98
74
41
27
Bayesian: Adjustedc
0.5
67
30
98
79
65
66
0.7
55
15
98
72
41
57
cEqual odds = 0.33 each eye color category, adjusted odds = 0.39 brown, 0.44 blue, 0.17 intermediate.
Fig. 2Corrected frequency of overall correct, incorrect, and inconclusive eye color predictions using the multinomial regression model. a) Predictions under IrisPlex parameters at the 0.5 threshold, b) predictions under adjusted parameters at the 0.5 threshold, c) predictions under IrisPlex parameters at the 0.7 threshold, and d) predictions under adjusted parameters at the 0.7 threshold.
Fig. 3The corrected frequencies of overall correct, incorrect, and inconclusive eye color predictions using the Bayesian model. a) Predictions under equal odds at the 0.5 threshold, b) predictions under adjusted frequency odds at the 0.5 threshold, c) predictions under equal odds at the 0.7 threshold, and d) predictions under adjusted frequency odds at the 0.7 threshold.
Table 4Corrected AUC values of each prediction model evaluating the training set (N = 100). AUC reflects model performance (ability to make accurate predictions). Higher AUC value indicates better model performance.
Table 5Corrected prediction model performance test characteristics (%) of both regression and Bayesian parameter sets after analysis of the training set (N = 100).
Model
Test Characteristics
Original
Corrected
Blue
Intermediate
Brown
Blue
Intermediate
Brown
MLR: IrisPlex
Sensitivity
95
93
84
95
94
87
Specificity
91
41
85
91
29
85
PPV
93
54
77
93
50
80
NPV
93
89
89
93
87
90
MLR: Adjusted
Sensitivity
95
95
93
96
98
89
Specificity
86
64
85
86
6
82
PPV
93
73
89
95
33
82
NPV
90
93
91
90
83
89
Bayesian: Equal odds
Sensitivity
93
92
97
95
0
92
Specificity
91
76
77
89
100
79
PPV
91
65
94
93
0
86
NPV
93
95
87
91
83
88
Bayesian: Adjusted
Sensitivity
93
93
87
95
0
92
Specificity
93
41
82
89
100
79
PPV
91
54
80
93
0
86
NPV
95
89
88
91
83
88
PPV = positive prediction value (correctly predicted positives), NPV = negative prediction value (correctly predicted negatives).
DNA phenotyping is a rapidly developing area of research in forensic biology. Externally visible characteristics (EVCs) can be determined based on genotype data, specifically based on single nucleotide polymorphisms (SNPs). These SNPs are chosen based on their association with genes related to the phenotypic expression of interest, with known examples in eye, hair, and skin color traits. DNA phenotyping has forensic importance when unknown biological samples at a crime scene do not result in a criminal database hit; a phenotypic profile of the sample can therefore be used to develop investigational leads.
30103-X&id=gr1.jpg){kind=link}
30103-X&id=gr2.jpg){kind=link}