Prediction of Eye Color from Genetic Data Using Bayesian Approach*

Predicting Physical Appearance from DNA Data-Towards Genomic Solutions

The idea of forensic DNA intelligence is to extract from genomic data any information that can help guide the investigation. The clues to the externally visible phenotype are of particular practical importance. The high heritability of the physical phenotype suggests that genetic data can be easily predicted, but this has only become possible with less polygenic traits. The forensic community has developed DNA-based predictive tools by employing a limited number of the most important markers analysed with targeted massive parallel sequencing. The complexity of the genetics of many other appearance phenotypes requires big data coupled with sophisticated machine learning methods to develop accurate genomic predictors. A significant challenge in developing universal genomic predictive methods will be the collection of sufficiently large data sets. These should be created using whole-genome sequencing technology to enable the identification of rare DNA variants implicated in phenotype determination. It is worth noting that the correctness of the forensic sketch generated from the DNA data depends on the inclusion of an age factor. This, however, can be predicted by analysing epigenetic data. An important limitation preventing whole-genome approaches from being commonly used in forensics is the slow progress in the development and implementation of high-throughput, low DNA input sequencing technologies. The example of palaeoanthropology suggests that such methods may possibly be developed in forensics.

Eye color prediction using the IrisPlex system: a limited pilot study in the Iraqi population

Background

Forensic DNA phenotyping has gained momentum in the recent past due to the prediction of externally visible characters (EVCs) from the biological sample. The most common phenotypes like eye, hair, and skin color are predicted from the biological samples using a web-based system called IrisPlex. Based on six genetic SNPs, the IrisPlex system is developed and validated for its prediction accuracy in diverse ethnic groups worldwide. In previous studies, this system proved to have significant prediction accuracy. The EVCs vary substantially based on different geographical locations. Hence, the objective of this study was to validate the accuracy of the IrisPlex system in predicting the eye colors in the Iraqi population.

Methods

Six genetic single-nucleotide polymorphisms SNPs (HERC2-rs12913832, OCA2- rs1800407, SLC24A4-rs12896399, SLC45A2- rs16891982, TYR-rs1393350, and IRF4- rs12203592) in 58 Iraqi subjects were performed using Sequenom MassARRAY Genotyping. According to Liu et al., a predicted probability of 0.7 was considered as the threshold.

Results

Participants in this study of brown color were observed in 44.83%, intermediate in 43.1%, and blue in 12.07%. Completely predictive accuracy is obtained in 1; we observed the AUC at threshold 0.7 was 0.91 for brown, 0.79 for blue, and 0.60 for intermediate eye color. The sensitivity was 42.85% for blue, 0% for intermediate eye color, and 100% for brown-colored eye. Specificity was 100% for blue, 100% for intermediate, and 78.13% for brown eye color.

Conclusion

Hence, it was concluded that the prediction accuracy of the IrisPlex system for blue and brown color eye in the Iraqi population is significant in the studied population size. However, a pivotal study with larger sample size is required to represent the prediction accuracy of the IrisPlex system in the whole Iraqi population.

Exploring the possibility of predicting human head hair greying from DNA using whole-exome and targeted NGS data

BACKGROUND

Greying of the hair is an obvious sign of human aging. In addition to age, sex- and ancestry-specific patterns of hair greying are also observed and the progression of greying may be affected by environmental factors. However, little is known about the genetic control of this process. This study aimed to assess the potential of genetic data to predict hair greying in a population of nearly 1000 individuals from Poland.

RESULTS

The study involved whole-exome sequencing followed by targeted analysis of 378 exome-wide and literature-based selected SNPs. For the selection of predictors, the minimum redundancy maximum relevance (mRMRe) method was used, and then two prediction models were developed. The models included age, sex and 13 unique SNPs. Two SNPs of the highest mRMRe score included whole-exome identified KIF1A rs59733750 and previously linked with hair loss FGF5 rs7680591. The model for greying vs. no greying prediction achieved accuracy of cross-validated AUC = 0.873. In the 3-grade classification cross-validated AUC equalled 0.864 for no greying, 0.791 for mild greying and 0.875 for severe greying. Although these values present fairly accurate prediction, most of the prediction information was brought by age alone. Genetic variants explained < 10% of hair greying variation and the impact of particular SNPs on prediction accuracy was found to be small.

CONCLUSIONS

The rate of changes in human progressive traits shows inter-individual variation, therefore they are perceived as biomarkers of the biological age of the organism. The knowledge on the mechanisms underlying phenotypic aging can be of special interest to the medicine, cosmetics industry and forensics. Our study improves the knowledge on the genetics underlying hair greying processes, presents prototype models for prediction and proves hair greying being genetically a very complex trait. Finally, we propose a four-step approach based on genetic and epigenetic data analysis allowing for i) sex determination; ii) genetic ancestry inference; iii) greying-associated SNPs assignment and iv) epigenetic age estimation, all needed for a final prediction of greying.

Non-CYP2D6 Variants Selected by a GWAS Improve the Prediction of Impaired Tamoxifen Metabolism in Patients with Breast Cancer

A certain minimum plasma concentration of (Z)-endoxifen is presumably required for breast cancer patients to benefit from tamoxifen therapy. In this study, we searched for DNA variants that could aid in the prediction of risk for insufficient (Z)-endoxifen exposure. A metabolic ratio (MR) corresponding to the (Z)-endoxifen efficacy threshold level was adopted as a cutoff value for a genome-wide association study comprised of 287 breast cancer patients. Multivariate regression was used to preselect variables exhibiting an independent impact on the MR and develop models to predict below-threshold MR values. In total, 15 single-nucleotide polymorphisms (SNPs) were significantly associated with below-threshold MR values. The strongest association was with rs8138080 (WBP2NL). Two alternative models for MR prediction were developed. The predictive accuracy of Model 1, including rs7245, rs6950784, rs1320308, and the CYP2D6 genotype, was considerably higher than that of the CYP2D6 genotype alone (AUC 0.879 vs 0.758). Model 2, which was developed using the same three SNPs as for Model 1 plus rs8138080, appeared as an interesting alternative to the full CYP2D6 genotype testing. In conclusion, the four novel SNPs, tested alone or in combination with the CYP2D6 genotype, improved the prediction of impaired tamoxifen-to-endoxifen metabolism, potentially allowing for treatment optimization.

Towards broadening Forensic DNA Phenotyping beyond pigmentation: Improving the prediction of head hair shape from DNA

Human head hair shape, commonly classified as straight, wavy, curly or frizzy, is an attractive target for Forensic DNA Phenotyping and other applications of human appearance prediction from DNA such as in paleogenetics. The genetic knowledge underlying head hair shape variation was recently improved by the outcome of a series of genome-wide association and replication studies in a total of 26,964 subjects, highlighting 12 loci of which 8 were novel and introducing a prediction model for Europeans based on 14 SNPs. In the present study, we evaluated the capacity of DNA-based head hair shape prediction by investigating an extended set of candidate SNP predictors and by using an independent set of samples for model validation. Prediction model building was carried out in 9674 subjects (6068 from Europe, 2899 from Asia and 707 of admixed European and Asian ancestries), used previously, by considering a novel list of 90 candidate SNPs. For model validation, genotype and phenotype data were newly collected in 2415 independent subjects (2138 Europeans and 277 non-Europeans) by applying two targeted massively parallel sequencing platforms, Ion Torrent PGM and MiSeq, or the MassARRAY platform. A binomial model was developed to predict straight vs. non-straight hair based on 32 SNPs from 26 genetic loci we identified as significantly contributing to the model. This model achieved prediction accuracies, expressed as AUC, of 0.664 in Europeans and 0.789 in non-Europeans; the statistically significant difference was explained mostly by the effect of one EDAR SNP in non-Europeans. Considering sex and age, in addition to the SNPs, slightly and insignificantly increased the prediction accuracies (AUC of 0.680 and 0.800, respectively). Based on the sample size and candidate DNA markers investigated, this study provides the most robust, validated, and accurate statistical prediction models and SNP predictor marker sets currently available for predicting head hair shape from DNA, providing the next step towards broadening Forensic DNA Phenotyping beyond pigmentation traits.

Further evidence for population specific differences in the effect of DNA markers and gender on eye colour prediction in forensics

The genetics of eye colour has been extensively studied over the past few years, and the identified polymorphisms have been applied with marked success in the field of Forensic DNA Phenotyping. A picture that arises from evaluation of the currently available eye colour prediction markers shows that only the analysis of HERC2-OCA2 complex has similar effectiveness in different populations, while the predictive potential of other loci may vary significantly. Moreover, the role of gender in the explanation of human eye colour variation should not be neglected in some populations. In the present study, we re-investigated the data for 1020 Polish individuals and using neural networks and logistic regression methods explored predictive capacity of IrisPlex SNPs and gender in this population sample. In general, neural networks provided higher prediction accuracy comparing to logistic regression (AUC increase by 0.02–0.06). Four out of six IrisPlex SNPs were associated with eye colour in the studied population. HERC2 rs12913832, OCA2 rs1800407 and SLC24A4 rs12896399 were found to be the most important eye colour predictors (p < 0.007) while the effect of rs16891982 in SLC45A2 was less significant. Gender was found to be significantly associated with eye colour with males having ~1.5 higher odds for blue eye colour comparing to females (p = 0.002) and was ranked as the third most important factor in blue/non-blue eye colour determination. However, the implementation of gender into the developed prediction models had marginal and ambiguous impact on the overall accuracy of prediction confirming that the effect of gender on eye colour in this population is small. Our study indicated the advantage of neural networks in prediction modeling in forensics and provided additional evidence for population specific differences in the predictive importance of the IrisPlex SNPs and gender.

A comparison of genomic profiles of complex diseases under different models

Background

Various approaches are being used to predict individual risk to polygenic diseases from data provided by genome-wide association studies. As there are substantial differences between the diseases investigated, the data sets used and the way they are tested, it is difficult to assess which models are more suitable for this task.

Results

We compared different approaches for seven complex diseases provided by the Wellcome Trust Case Control Consortium (WTCCC) under a within-study validation approach. Risk models were inferred using a variety of learning machines and assumptions about the underlying genetic model, including a haplotype-based approach with different haplotype lengths and different thresholds in association levels to choose loci as part of the predictive model. In accordance with previous work, our results generally showed low accuracy considering disease heritability and population prevalence. However, the boosting algorithm returned a predictive area under the ROC curve (AUC) of 0.8805 for Type 1 diabetes (T1D) and 0.8087 for rheumatoid arthritis, both clearly over the AUC obtained by other approaches and over 0.75, which is the minimum required for a disease to be successfully tested on a sample at risk, which means that boosting is a promising approach. Its good performance seems to be related to its robustness to redundant data, as in the case of genome-wide data sets due to linkage disequilibrium.

Conclusions

In view of our results, the boosting approach may be suitable for modeling individual predisposition to Type 1 diabetes and rheumatoid arthritis based on genome-wide data and should be considered for more in-depth research.

Electronic supplementary material

The online version of this article (doi:10.1186/s12920-015-0157-2) contains supplementary material, which is available to authorized users.

Evaluation of the predictive capacity of DNA variants associated with straight hair in Europeans

DNA-based prediction of hair morphology, defined as straight, curly or wavy hair, could contribute to an improved description of an unknown offender and allow more accurate forensic reconstructions of physical appearance in the field of forensic DNA phenotyping. Differences in scalp hair morphology are significant at the worldwide scale and within Europe. The only genome-wide association study made to date revealed the Trichohyalin gene (TCHH) to be significantly associated with hair morphology in Europeans and reported weaker associations for WNT10A and FRAS1 genes. We conducted a study that centered on six SNPs located in these three genes with a sample of 528 individuals from Poland. The predictive capacity of the candidate DNA variants was evaluated using logistic regression; classification and regression trees; and neural networks, by applying a 10-fold cross validation procedure. Additionally, an independent test set of 142 males from six European populations was used to verify performance of the developed prediction models. Our study confirmed association of rs11803731 (TCHH), rs7349332 (WNT10A) and rs1268789 (FRAS1) SNPs with hair morphology. The combined genotype risk score for straight hair had an odds ratio of 2.7 and these predictors explained ∼ 8.2% of the total variance. The selected three SNPs were found to predict straight hair with a high sensitivity but low specificity when a 10-fold cross validation procedure was applied and the best results were obtained using the neural networks approach (AUC=0.688, sensitivity=91.2%, specificity=23.0%). Application of the neural networks model with 65% probability threshold on an additional test set gave high sensitivity (81.4%) and improved specificity (50.0%) with a total of 78.7% correct calls, but a high non-classification rate (66.9%). The combined TTGGGG SNP genotype for rs11803731, rs7349332, rs1268789 (European frequency=4.5%) of all six straight hair-associated alleles was identified as the best predictor, giving >80% probability of straight hair. Finally, association testing of 44 SNPs previously identified to be associated with male pattern baldness revealed a suggestive association with hair morphology for rs4679955 on 3q25.1. The study results reported provide the starting point for the development of a predictive test for hair morphology in Europeans. More studies are now needed to discover additional determinants of hair morphology to improve the predictive accuracy of this trait in forensic analysis.

Quantitative assessment of skin, hair, and iris variation in a diverse sample of individuals and associated genetic variation

OBJECTIVES

The main goals of this study are to 1) quantitatively measure skin, hair, and iris pigmentation in a diverse sample of individuals, 2) describe variation within and between these samples, and 3) demonstrate how quantitative measures can facilitate genotype-phenotype association tests.

MATERIALS AND METHODS

We quantitatively characterize skin, hair, and iris pigmentation using the Melanin (M) Index (skin) and CIELab values (hair) in 1,450 individuals who self-identify as African American, East Asian, European, Hispanic, or South Asian. We also quantify iris pigmentation in a subset of these individuals using CIELab values from high-resolution iris photographs. We compare mean skin M index and hair and iris CIELab values among populations using ANOVA and MANOVA respectively and test for genotype-phenotype associations in the European sample.

RESULTS

All five populations are significantly different for skin (P <2 × 10(-16) ) and hair color (P <2 × 10(-16) ). Our quantitative analysis of iris and hair pigmentation reinforces the continuous, rather than discrete, nature of these traits. We confirm the association of three loci (rs16891982, rs12203592, and rs12913832) with skin pigmentation and four loci (rs12913832, rs12203592, rs12896399, and rs16891982) with hair pigmentation. Interestingly, the derived rs12203592 T allele located within the IRF4 gene is associated with lighter skin but darker hair color.

DISCUSSION

The quantitative methods used here provide a fine-scale assessment of pigmentation phenotype and facilitate genotype-phenotype associations, even with relatively small sample sizes. This represents an important expansion of current investigations into pigmentation phenotype and associated genetic variation by including non-European and admixed populations. Am J Phys Anthropol 160:570-581, 2016. © 2015 Wiley Periodicals, Inc.

Evaluation of DNA variants associated with androgenetic alopecia and their potential to predict male pattern baldness

Androgenetic alopecia, known in men as male pattern baldness (MPB), is a very conspicuous condition that is particularly frequent among European men and thus contributes markedly to variation in physical appearance traits amongst Europeans. Recent studies have revealed multiple genes and polymorphisms to be associated with susceptibility to MPB. In this study, 50 candidate SNPs for androgenetic alopecia were analyzed in order to verify their potential to predict MPB. Significant associations were confirmed for 29 SNPs from chromosomes X, 1, 5, 7, 18 and 20. A simple 5-SNP prediction model and an extended 20-SNP model were developed based on a discovery panel of 305 males from various European populations fitting one of two distinct phenotype categories. The first category consisted of men below 50 years of age with significant baldness and the second; men aged 50 years or older lacking baldness. The simple model comprised the five best predictors: rs5919324 near AR, rs1998076 in the 20p11 region, rs929626 in EBF1, rs12565727 in TARDBP and rs756853 in HDAC9. The extended prediction model added 15 SNPs from five genomic regions that improved overall prevalence-adjusted predictive accuracy measured by area under the receiver characteristic operating curve (AUC). Both models were evaluated for predictive accuracy using a test set of 300 males reflecting the general European population. Applying a 65% probability threshold, high prediction sensitivity of 87.1% but low specificity of 42.4% was obtained in men aged <50 years. In men aged ≥50, prediction sensitivity was slightly lower at 67.7% while specificity reached 90%. Overall, the AUC=0.761 calculated for men at or above 50 years of age indicates these SNPs offer considerable potential for the application of genetic tests to predict MPB patterns, adding a highly informative predictive system to the emerging field of forensic analysis of externally visible characteristics.

Assay Development and Validation of an 8-SNP Multiplex Test to Predict Eye and Skin Coloration

Identifying human remains is one of the many responsibilities of forensic scientists. An eye- and skin-color predictor translates genotypic information into phenotypic description. Eight single nucleotide polymorphisms (SNPs) are utilized for this predictor, five for eye, and six for skin coloration. Here, we describe the development and validation of an 8-SNP multiplex assay that consists of a multiplex PCR, followed by a multiplexed single-base primer extension reaction generating fluorescently labeled oligonucleotides of distinct length that are detected by multicolor capillary electrophoresis. Validation of this assay included tests for reproducibility, reliability, sensitivity, species specificity, its performance on degraded DNA, and on forensic samples. It can be concluded that the 8-SNP multiplex assay is robust and can be used on challenging samples, including bones, to reliably determine the genotypes to predict eye and skin color of individuals. This information can assist in the identification of human remains and missing persons.

Evaluation of the IrisPlex DNA-based eye color prediction assay in a United States population

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. IrisPlex, an eye color prediction assay, has previously shown high prediction rates for blue and brown eye color in a Dutch European population. The objective of this work was to evaluate its utility in a North American population. We evaluated six SNPs included in the IrisPlex assay in population sample collected from a USA college campus. We used a quantitative method of eye color classification based on (RGB) color components of digital photographs of the eye taken from each study volunteer so that each eye was placed in one of three eye color categories: brown, intermediate, or blue. Objective color classification was shown to correlate with basic human visual determination making it a feasible option for use in future prediction assay development. Using these samples and various models, the maximum prediction accuracies of the IrisPlex system after allele frequency adjustment was 58% and 95% brown and blue eye color predictions, respectively, and 11% for intermediate eye colors. Future developments should include incorporation of additional informative SNPs, specifically related to the intermediate eye color, and we recommend the use of a Bayesian approach as a prediction model as likelihood ratios can be determined for reporting purposes.

Prediction of eye color in the Slovenian population using the IrisPlex SNPs

AIM

To evaluate the accuracy of eye color prediction based on six IrisPlex single nucleotide polymorphisms (SNP) in a Slovenian population sample.

METHODS

Six IrisPlex predictor SNPs (HERC2 - rs12913832, OCA2 - rs1800407, SLC45A2 - rs16891982 and TYR - rs1393350, SLC24A4 - rs12896399, and IRF4 - rs12203592) of 105 individuals were analyzed using single base extension approach and SNaPshot chemistry. The IrisPlex multinomial regression prediction model was used to infer eye color probabilities. The accuracy of the IrisPlex was assessed through the calculation of sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and the area under the receiver characteristic operating curves (AUC).

RESULTS

Blue eye color was observed in 44.7%, brown in 29.6%, and intermediate in 25.7% participants. Prediction accuracy expressed by the AUC was 0.966 for blue, 0.913 for brown, and 0.796 for intermediate eye color. Sensitivity was 93.6% for blue, 58.1% for brown, and 0% for intermediate eye color. Specificity was 93.1% for blue, 89.2% for brown, and 100% for intermediate eye color. PPV was 91.7% for blue and 69.2% for brown color. NPV was 94.7% for blue and 83.5% for brown eye color. These values indicate prediction accuracy comparable to that established in other studies.

CONCLUSION

Blue and brown eye color can be reliably predicted from DNA samples using only six polymorphisms, while intermediate eye color defies prediction, indicating that more research is needed to genetically predict the whole variation of eye color in humans.

Insulin resistance and inflammation predict kinetic body weight changes in response to dietary weight loss and maintenance in overweight and obese subjects by using a Bayesian network approach

BACKGROUND

The ability to identify obese subjects who will lose weight in response to energy restriction is an important strategy in obesity treatment.

OBJECTIVE

We aimed to identify obese subjects who would lose weight and maintain weight loss through 6 wk of energy restriction and 6 wk of weight maintenance.

DESIGN

Fifty obese or overweight subjects underwent a 6-wk energy-restricted, high-protein diet followed by another 6 wk of weight maintenance. Network modeling by using combined biological, gut microbiota, and environmental factors was performed to identify predictors of weight trajectories.

RESULTS

On the basis of body weight trajectories, 3 subject clusters were identified. Clusters A and B lost more weight during energy restriction. During the stabilization phase, cluster A continued to lose weight, whereas cluster B remained stable. Cluster C lost less and rapidly regained weight during the stabilization period. At baseline, cluster C had the highest plasma insulin, interleukin (IL)-6, adipose tissue inflammation (HAM56+ cells), and Lactobacillus/Leuconostoc/Pediococcus numbers in fecal samples. Weight regain after energy restriction correlated positively with insulin resistance (homeostasis model assessment of insulin resistance: r = 0.5, P = 0.0002) and inflammatory markers (IL-6; r = 0.43, P = 0.002) at baseline. The Bayesian network identified plasma insulin, IL-6, leukocyte number, and adipose tissue (HAM56) at baseline as predictors that were sufficient to characterize the 3 clusters. The prediction accuracy reached 75.5%.

CONCLUSION

The resistance to weight loss and proneness to weight regain could be predicted by the combination of high plasma insulin and inflammatory markers before dietary intervention.

Improved eye- and skin-color prediction based on 8 SNPs

Aim

To improve the 7-plex system to predict eye and skin color by increasing precision and detailed phenotypic descriptions.

Methods

Analysis of an eighth single nucleotide polymorphism (SNP), rs12896399 (SLC24A4), showed a statistically significant association with human eye color (P = 0.007) but a rather poor strength of agreement (κ = 0.063). This SNP was added to the 7-plex system (rs12913832 at HERC2, rs1545397 at OCA2, rs16891982 at SLC45A2, rs1426654 at SLC24A5, rs885479 at MC1R, rs6119471 at ASIP, and rs12203592 at IRF4). Further, the instruction guidelines on the interpretation of genotypes were changed to create a new 8-plex system. This was based on the analysis of an 803-sample training set of various populations. The newly developed 8-plex system can predict the eye colors brown, green, and blue, and skin colors light, not dark, and not light. It is superior to the 7-plex system with its additional ability to predict blue eye and light skin color.

Results

The 8-plex system was tested on an additional 212 samples, the test set. Analysis showed that the number of positive descriptions for eye colors as being brown, green, or blue increased significantly (P = 6.98e-15, z-score: -7.786). The error rate for eye-color prediction was low, at approximately 5%, while the skin color prediction showed no error in the test set (1% in training set).

Conclusions

We can conclude that the new 8-plex system for the prediction of eye and skin color substantially enhances its former version.

A single-nucleotide polymorphism (SNP) multiplex system: the association of five SNPs with human eye and hair color in the Slovenian population and comparison using a Bayesian network and logistic regression model

AIM

To analyze two phenotype characteristics--eye and hair color--using single-nucleotide polymorphisms (SNPs) and evaluate their prediction accuracy in Slovenian population.

METHODS

Twelve SNPs (OCA2 - rs1667394, rs7170989, rs1800407, rs7495174; HERC2 - rs1129038, rs12913832; MC1R - rs1805005, rs1805008; TYR - rs1393350; SLC45A2 - rs16891982, rs26722; SLC24A5 - rs1426654) were used for the development of a single multiplex assay. The single multiplex assay was based on SNaPshot chemistry and capillary electrophoresis. In order to evaluate the accuracy of the prediction of eye and hair color, we used the logistic regression model and the Bayesian network model, and compared the parameters of both.

RESULTS

The new single multiplex assay displayed high levels of genotyping sensitivity with complete profiles generated from as little as 62 pg of DNA. Based on a prior evaluation of all SNPs in a single multiplex, we focused on the five most statistically significant in our population in order to investigate the predictive value. The two prediction models performed reliably without prior ancestry information, and revealed very good accuracy for both eye and hair color. Both models determined the highest predictive value for rs12913832 (P<0.0001), while the other four SNPs (rs1393350, rs1800407, rs1805008, and rs7495174) showed additional association for color prediction.

CONCLUSION

We developed a sensitive and reliable single multiplex genotyping assay. More samples from different populations should be analyzed before this assay could be used as one of the supplemental tools in tracing unknown individuals in more complicated crime investigations.

SNP model development for the prediction of eye colour in New Zealand

The ability to predict externally visible characteristics (EVCs) from DNA has appeal for use in forensic science, particularly where a forensic database match is not made and an eye witness account is unavailable. This technology has yet to be implemented in casework in New Zealand. The broad cultural diversity and likely population stratification within New Zealand dictates that any EVC predictions made using anonymous DNA must perform accurately in the absence of knowledge of the donor's ancestral background. Here we construct classification tree models with SNPs of known association with eye colour phenotypes in three categories, blue vs. non-blue, brown vs. non-brown and intermediate vs. non-intermediate. A set of nineteen SNPs from ten different known or suspected pigmentation genes were selected from the literature. A training dataset of 101 unrelated individuals from the New Zealand population and representing different ancestral backgrounds were used. We constructed four alternate models capable of predicting eye colour from the DNA genotypes of SNPs located within the HERC2, OCA2, TYR and SLC24A4 genes using probability calculation and classification trees. The final model selected for eye colour prediction exhibited high levels of accuracy for both blue (89%) and brown eye colour (94%). Models were further assessed with a test set of 25 'blind' samples where phenotype was unknown, with blue and brown eye colour predicted correctly where model thresholds were met. Classification trees offer an aesthetically simple and comprehendible model to predict blue and brown eye colour.