The more the merrier? How a few SNPs predict pigmentation phenotypes in the
Northern German population
Amke Caliebe1,*, Melanie Harder2*, Rebecca Schuett2,4, Michael Krawczak1, Almut Nebel3, Nicole von Wurmb-Schwark2
1
Institute of Medical Informatics and Statistics, 2Institute of Legal Medicine, 3Institute of Clinical Molecular Biology, all at
Christian-Albrechts University Kiel, Germany; 4current address: State Criminal Investigation Department of Lower Saxony
Correspondence to:
PD Dr. rer. nat. Nicole von Wurmb-Schwark, Institute of Legal Medicine, Christian-Albrechts University Kiel, University
Hospital Schleswig-Holstein, Arnold-Heller Str. 12, 24105 Kiel, tel.: +49 431 597 3633, fax: +49 431 597 3612, mail:
nvonwurmb@rechtsmedizin.uni-kiel.de
*
These authors contributed equally to this work.
Material and Methods
Study population
A total of 400 unrelated individuals (197 male, 203 female) from Northern Germany were recruited for
our study between 2010 and 2011. The median age was 27 years (inter-quartile range: 24-33 years). All
individuals were born in Germany and had German parents and grandparents (self-report). All 400
participants were recruited and investigated in the same way. Whilst the first 300 participants were
included in the ‘modelling sample’ of stage 1 (used for SNP selection), the remaining 100 individuals
constituted the ‘estimation sample’ of stage 2 (prediction evaluation). All participants gave written
informed consent prior to the study. Genotype and phenotype data were de-identified for analysis purposes
according to the declaration of Helsinki. The project was approved by the Ethics Committee of the
Medical Faculty of Christian-Albrechts University Kiel.
Phenotyping
Pigmentation phenotypes were documented by photographs taken at daylight conditions and from a
distance of 30 cm, using a Canon EOS 400D (18-55 mm focal length). For each participant, one
photograph was taken of each eye, the scalp hair and the inner arm. Photographs were normalised using
the standard functions in Photoshop 4.0, and consensus phenotype calling was carried out by two raters by
discussion. In the rare cases where no agreement was reached a third party was involved.
Eye colour was divided into three categories, namely blue, green and brown (Fig. 1a). Individual skin type
was classified applying the Fitzpatrick scheme (1988) to the inner arm. Hair colour type was defined in
multi-tiered fashion. To this end, a collection of coloured hair strands obtained from a hairdresser was
categorised into nine evenly graded types of shading, ranging from light blond (type I) to black (type IX)
(Fig. 1b). Strands of red hair or with red tint were omitted from this classification because it was intended
to address the light-dark component only. Then, hair colour was divided into two sub-phenotypes, namely
the red tint component (yes/no) and the light-dark component (I-IX). For each individual, the presence of
red tint was ascertained (by questioning) in head hair, facial hair (beard), axillary hair or pubic hair. If an
individual had red head hair, this was noted separately to enable a separate analysis for this special
phenotype. The light-dark component was determined by reference to the hair strand collection mentioned
above. Here, individuals with recognizable red tint were classified according to their basic hair colour
type. For example, strawberry blonds were deemed class I (blond) whereas people with auburn hair were
classified as one of IV, V or VI (brown). While this was possible for 22 red-haired individuals, four had
no definable basic hair colour. These were excluded from the analysis of the light-dark component. No
participants with exclusively white hair were included in the study. When a participant had dyed hair, the
original hair colour was determined by the hairline.
Genotyping
Buccal swabs (COPAN) were taken from all 400 participants and DNA was extracted using Chelex 100
(Walsh et al. 1991). In a comprehensive PubMed search, 12 SNPs were identified as promising candidates
for further analysis using the following criteria (Table 1, Supplementary Table S1): large odds ratio,
validation in several independent studies, large sample sizes and adequate population backgrounds, low to
no linkage disequilibrium with other candidate markers and suitability for genotyping in a single assay. In
addition to the 12 SNPs, participants were genotyped for rs1426654 (SLC24A5), rs1129038 (HERC2) and
rs1667394 (OCA2). SNP rs1426654 is a European ancestry marker (Giardina et al. 2008) used to control
population background. SNPs rs1129038 and rs1667394 served as genotyping quality markers because
they are in perfect LD with candidate SNPs rs12913832 and rs916977 respectively (Mengel-From et al.
2010; Sturm et al. 2008).
Primers were designed and checked for possible dimer and hairpin structures using the DNAstar
Lasergene v8.1.2 software and BLAST. PCR fragments had to be shorter than 200 bp in order to meet the
standards of reliable forensic or ancient DNA analysis.
For DNA amplification, a Multiplex PCR Master Mix (Qiagen) was used in a total reaction volume of
12.5 µl, with 0.2-0.5 ng template DNA. PCR was performed with a thermal cycler 2700 (Life
Technologies) under the following conditions: (1) 95° C 15 min, (2) 35 cycles of 94° C 30 sec, 64° C
(SNPs nos. 2-5, 7-8, 10 in Table 1) or 58° C (SNPs nos. 1, 6, 9, 11-12) 90 sec, 72° C 1 min and (3) 60° C
30 min. PCR products were purified using ExoSAP-IT (Affymetrix) according to the manufacturer’s
protocol. Single-base extension (SBE) was carried out in a total reaction volume of 7 µl, including 0.5 µl
of cleaned PCR products, using the SNaPshot Multiplex Kit (Life Technologies) on the same PCR cycler
as before. The SBE cycling conditions were as follows: 25 cycles of 96° C 10 sec, 55° C 5 sec, 60° C 30
sec. Fragment analysis was performed with the ABI Prism 3130 Genetic Analyzer (Life Technologies)
using GeneMapper v3.2. For more information on primer sequences and concentrations, see Table S2a.
The model proposed by Walsh et al. (2011) for the prediction of eye colour is based upon six SNPs. Five
of these had been genotyped in all our study participants before (Table 1, SNPs nos. 1, 3, 10-12). The 100
individuals of stage 2 were additionally genotyped for rs16891982 (SLC45A2) as described above, with an
annealing temperature of 58° C in the first PCR. For more information on primer sequences and
concentrations, see Table S2b.
The model devised by Branicki et al. (2011) for predicting hair colour is based upon 13 single or
compound markers. Three of these were also included in our set of candidate SNPs (Table 1, SNPs nos. 1,
3, 10) and were genotyped in all participants. The 100 stage 2 individuals were also genotyped for the
remaining 10 markers, namely two compound markers in MC1R and rs1042602 (TYR), rs4959270
(EXOC2), rs28777 (SLC45A2), rs683 (TYRP1), rs2402130 (SLC24A4), rs12821256 (KITLG), rs16891982
(SLC45A2) and rs2378249 (ASIP). SNPs were analyzed as described above, with an annealing
temperature of 58° C for the first PCR. The MC1R markers were analyzed by sequencing the whole locus.
To this end, a 1080 bp fragment was amplified and sequenced with an ABI Prism 3130xl Genetic
Analyzer using the Big Dye Terminator v3.1 Cycle Sequencing Kit ( both Life Technologies), following
the manufacturer’s protocols. See Table S2c for more information on primers used in this study.
Genotype and phenotype data of this study were submitted to the European Genome-phenome archive
(EGA, https://ega.crg.eu) with study accession number EGAS00001001174 (sample/proband ids
EGAN00001268626-EGAN00001269025).
Statistical analysis
Genotypes for all markers of the two previously published models (or marker sets) (Branicki et al. 2011;
Walsh et al. 2011) were only available for 100 individuals in our study (stage 2). To compare the
predictive capability of the two marker sets to a model derived specifically for our target population, the
300 individuals of stage 1 were used to detect significant genotype-phenotype associations and to create
an appropriate prediction model. Data from stage 2 then served for estimation and comparison of the
predictive capability (sensitivity, specificity, predictive accuracy, area under the receiver operating
characteristic curve AUC) for each new model and the two previously published models (Branicki et al.
2011; Walsh et al. 2011). For illustration, we also performed model selection and prediction evaluation on
the whole data set (i.e. stages 1 and 2 combined), using cross-validation to estimate sensitivity and
specificity.
Sample size calculations indicated that approximately 100 individuals per group would suffice to detect an
OR of 3 as nominally significant, depending upon the minor allele frequency of the SNP of interest, and
150 individuals per group after Bonferroni adjustment (12 SNPs tested, 80% power, 5% significance
level). Stage 1 therefore comprised 300 individuals. The association between a given trait (i.e. eye colour,
hair colour/red tint, hair colour/light-dark component, skin colour) and a candidate SNP was tested for
statistical significance using regression models. To allow for scarce genotypes, we performed permutation
tests (100,000 permutations) in addition to standard asymptotic tests. Since p values were not found to be
notably different, only p values from permutation tests will be given. Each SNP was analyzed both
individually (simple regression) and in combination with other candidate SNPs (multiple regression with
backward selection), also allowing for possible SNP-SNP interactions. Genotypic, additive allelic,
dominant and recessive models were considered for each SNP. Results, however, will be presented for the
additive model only because this model required the least parameters but yielded consistently large
effects. To derive robust prediction models, phenotypes were categorised in various ways. Dependent on
the scaling of the outcome, we performed logistic, linear, ordinal (proportional odds) and/or multinomial
logistic regression. Since blue was by far the most frequent eye colour in our study, the analysis of eye
colour was confined to the discrimination between blue and non-blue. For hair colour, red tint was treated
as a dichotomous outcome whereas the light-dark component was treated in three different ways, either as
dichotomous (blond vs. non-blond), ordinal, or quantitative (nine types of increasing darkness). Skin
colour was treated either as dichotomous (types I-II vs. types III-IV) or as ordinal. Model selection was
performed differently for the four traits. For eye colour and red tint, SNPs that remained significant in the
multiple logistic regression analysis after backward selection and adjustment for multiple testing were
included in the final model. For the light-dark component of hair colour, and for skin colour, a SNP had to
be significant in all or in all but one of the multiple regression analyses after backward selection using
different outcome definitions (i.e. at least two of three analyses for the light-dark component, at least one
of two analyses for skin colour).
The relationships between traits were analyzed using logistic regression analysis, treating one trait as the
dependent variable and the other traits as independent variables, both with and without the additional
inclusion of SNP genotypes. All four traits were encoded as dichotomous variables in these analyses.
Model selection was again performed by backward selection. Multidimensional scaling (MDS) was used
to detect and visualise patterns in the phenotype data.
The predictive capability of a derived model was evaluated by means of the phenotype probability  from
logistic regression analysis. This was done only for dichotomous outcomes (e.g. blue vs. non-blue eye
colour). If >0.5, the corresponding phenotype was assumed to be present. Predictive capability was
quantified by the sensitivity, specificity, predictive accuracy and AUC of the model in question.
All statistical analyses were performed with R v2.10.1 (R Development Core Team 2009) unless indicated
otherwise. Hardy-Weinberg equilibrium was assessed by means of the exact test implemented in R
package genetics (Warnes et al. 2008). Package MASS was used for ordinal and multinomial regression
(Venables and Ripley 2002). Permutation tests of the linear and logistic regression models were performed
with package glmperm (Werft et al. 2013). For ordinal regression models, permutation tests were
programmed in house. The predictive capabilities of different models were evaluated with packages
DiagnosisMed (Brasil 2010) and pROC (Robin et al. 2011). The proportion of phenotype heritability
explained by a given marker was calculated according to So et al.(So et al. 2011). Note that these
estimates apply to single markers and do not take into account the characteristics of the respective
regression models. Furthermore, these estimates refer to the liability scale and therefore tend to be higher
than on the observation scale.
Sample size calculations were performed with the GPower software v3.0.8. All tests were two-sided and a
p value smaller than 0.05 was considered nominally statistically significant. P values were adjusted for
multiple testing using the Bonferroni method.
Supplementary Table 1: Terms for PubMed search
Category 1
human
Category 2
Category 3
Category 4
pigmentation
genotype/genotyping
prediction
eye colour/color
single nucleotide
polymorphism/
SNP/SNPs
determination
skin colour/color
hair colour/color
red hair/red tint
curly hair
For the PubMed search were used:
Always: one word each of categories 1-3
Often: one word of category 4
Sometimes: one word of category 5
Furthermore articles were investigated that cited already retrieved articles.
Category 5
origin/ancestry/
ancestral
forensic
forensic phenotyping/
forensic DNA phenotyping
Irisplex
Supplementary Table 2: Genotyping information for all analyzed SNPs
Supplementary Table 2a: Genotyping information for the 12 analyzed candidate SNPs and 3 control SNPs
Multiplex I
gene
size
(bp)
SNP-ID
MC1R
174
NC_000016.10:g89919709C>T
rs1805007
rs1805008
NC_000016.10:g.89919736C>T
OCA2
167
HERC
181
OCA2
64
OCA2
254
IRF4
121
HERC2
75
TYR
73
SLC24A5
137
MC1R
143
HERC2
176
SLC24A4
114
OCA2
66
HERC2
163
rs4778138
NC_000015.10:g.28090674A>G
rs1667394
NC_000015.10:g.28285036C>T
rs7495174
NC_000015.10:g.28099092A>G
rs1800407
NC_000015.10:g.27985172C>T
rs12203592
NC_000006.11:g.396321C>T
rs916977
NC_000015.10:g.28268218T>C
rs1393350
NC_000011.10:g.89277878G>A
rs1426654
NC_000015.10:g.48134287A>G
rs1805009
NC_000016.10:g.89920138G>C
rs1129038
NC_000015.10:g.28111713C>T
rs12896399
NC_000014.9:g.92307319G>T
rs4778241
NC_000015.10:g.28093567A>C
rs12913832
NC_000015.10:g.28120472A>G
primer sequence
SNP
allele
SBE-primer sequence
tail
total
length
(bp)
C/G/T
TCTCCATCTTCTACGCACTG
(GACT)5
40
C/T
AGCATCGTGACCCTGCCG
(A)14
32
0.8
G/A
GTGAAAATATAACATATCAAAATTG
(GACT)6
49
0.2
G/A
CATTGTTTCTTTGTTTGTTTGGT
(A)7
30
0.2
G/A
(C/T)
AGGCAAGTTCCCCTAAAGGT
(A)5
25
0,4
G/A
AGGCATACCGGCTCTCCC
(GACT)9
54
0.4
G/A
(C/T)
ACTTTGGTGGGTAAAAGAAGG
(GACT)8
53
0.1
G/A
GTGCAGCCTTGGCCAGCCTTCT
(GACT)6
46
0.05
G/A
CCTCAGTCCCTTCTCTGCAAC
(GACT)9
57
0.2
G/A
(C/T)
CTGAACTGCCCGCTGCCATGAAAGTTG
(GACT)4
43
0.2
G/C
CTCATCATCTGCAATGCCATCATC
(GACT)7
52
0.1
G/A
TGAGCCAGGCAGCAGAGC
(A)9
27
0.1
G/T
CTTTAGGTCAGTATATTTTGGG
0.1
C/A
(G/T)
TGTTGGCTGGTAGTTGCAATT
(GACT)6
45
0.05
G/A
GCCAGTTTCATTTGAGCATTAA
(A)13
35
[primer]
in µM
forward:GCCGTGGACCGCTACATC
0.6
reverse: GAAGAAGACCACGAGGCACAG
forward: GGAAAATCTGCACACTTAGAAA
reverse: GCTGTAAATTTCCTCCCATCA
forward: TTGGCAGCTTTTCTGTCTTCT
reverse: AAAATGAGAACTTGGTCAATCC
forward: GGCTCCGTCGCACCCGTCTG
reverse: GCGGCTTAGGAAGCAAGGCAAG
forward: CAGAGGTGCTTTGCGTACCTTATGGT
reverse: GGGGTAATGTTAGTTTGGCTCCCTGTTCTTA
forward: TCATGTGAAACCACAGGGCA
reverse: CTGGCACCAAAAGTACCACA
forward: CACAGTGGGGATGCAGTTTGAGTA
reverse: TTGGCCTTTCTGTTCTTCTTGACC
Multiplex II
forward: TATCCACCAACTCCTACTCTT
reverse:TTATCATTTGTAAAAGACCACAC
forward: GAAGAAAATAAAAATCACACTGAGTAAGC
reverse:CCTTGGATTGTCTCAGGATGTTGC
forward: CACCGCGCTCACCAGGAGC
reverse:GGCTGCATCTTCAAGAACTTCAACC
forward: GCCGACGACAGCAGCGACGAT
reverse:CAGACACACCAGGCAGCCTACAGTCT
forward: ATTGAGTATCCTATATTTTATCTG
reverse:TCTTGATGTTGTATTGATGAGG
forward: CTGGAAAGCAGTTTGACAGTT
reverse:GTGCAATTGTTGGCTGGTAG
forward: AAGAGGCGAGGCCAGTTTC
reverse:AGAAACGACAAGTAGACCATTT
reverse SBE-primer
22
Supplementary Table 2b: Genotyping information for additional SNP rs16891982 (SLC45A2) from Walsh et al. 30
Singelplex-PCR
gene
size
(bp)
SLC45A2
107
SNP-ID
rs16891982
NC_000005.10:g.33951588C>G
primer sequence
forward: AGAAACTTTTAGAAGACATCCTTAGGAGAGAGAAA
reverse: AAGAGGAGTCGAGGTTGGATGTTGG
[primer]
in µM
SNP
allele
SBE-primer sequence
tail
total
length
(bp)
0.2
C/G
(G/C)
AGGTTGGATGTTGGGGCTT
(GACT)7
47
reverse SBE-primer
Supplementary Table 2c: Genotyping information for additional single or compound markers from Branicki et al. 29
Multiplex-PCR
gene
size
(bp)
TYR
120
EXOC2
113
SLC45A2
109
SNP-ID
rs1042602
NC_000011.10:g.89178528C>A
rs4959270
NC_000006.12:g.457748C>A
rs28777
NC_000005.10:g.33958854C>A
primer sequence
forward: ATGACCTCTTTGTCTGGATGC
reverse: CTATGCCAAGGCAGAAAAGC
forward: CTGGGGTTTACGATTCAACA
reverse: AGGATGGAAAAGAACCACCA
forward: GTGGGAGTTCCATGCCTTT
reverse: TCCAAGAGTCGCATAGGACA
[primer]
in µM
SNP
allele
SBE-primer sequence
tail
total
length
(bp)
0.1
A/C
(T/G)
CAATGTCTCTCCAGATTTCA
(GACT)9
32
0.2
A/C
CCAAACTATGACACTATG
(GACT)
22
0.2
A/C
CATGTGATCCTCACAGCAG
(GACT)2
29
0.2
A/G
CATACGGAGCCCGTG
(GACT)7
43
0.1
C/T
(G/A)
GGGCATGTTACTACGGCAC
(GACT)5
39
0.2
A/G
CTAGGAACTACTTTGCACAGTA
(GACT)3
34
A/C
(T/G)
GCCTAGAACTTTAAT
(GACT)3
27
Duplex-PCR
SLC24A4
152
KITLG
106
ASIP
135
TYRP1
120
gene
size
(bp)
MC1R
1080
rs2402130
NC_000014.9:g.92334859G>A
rs12821256
NC_000012.12:g.88934558T>C
rs2378249
NC_000020.11:g.34630286G>A
rs683
NC_000009.12:g.12709305C>A
forward: ACCTGTCTCACAGTGCTGCT
reverse: TTCACCTCGATGACGATGAT
forward: TTAAGCTCTGTGTTTAGGGTTTTT
reverse: TGAGTCATGAGTGCTTTGTTCC
Singleplex-PCR
forward: GACCTCAGTTCTGGAGAAAGC
reverse: AAGGTGGCTGGTTTCAGTCT
Singleplex-PCR
forward: TTTTCTTTCACTTTATTACCTTCTTTC
0.2
reverse: AAAGATTCTGAAAGGGTCTTCC
Singleplex-PCR MC1R Exon
primer sequence
forward: GCAGCACCATGAACTAAGCA
reverse: TGCCCAGCACACTTAAAGC
concentration of each primer (µM)
0.4, 3.2 in the sequencing reaction
Supplementary Table 3: Genotype-phenotype association analysis of eye colour
Supplementary Table 3a: Association between SNP genotype and blue eye colour (stage 1)
Alleles
ref.
fav.
A
G
A
G
G
A
G
A
G
A
A
C
-
SNP-ID (gene)
rs12913832 (HERC2)
rs916977(HERC2)
rs1800407(OCA2)
rs7495174(OCA2)
rs4778138(OCA2)
rs4778241(OCA2)
rs1805007(MC1R)
rs1805008(MC1R)
rs1805009 (MC1R)
rs12203592(IRF4)
rs12896399 (SLC24A4)
rs1393350 (TYR)
Regression p value
simple
multiplea
<1.010-5
<1.010-5
n.s.
<1.010-5
<1.010-5
<1.010-5
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
<1.010-5 (1.210-4)
n.s.
0.0014 (0.017)
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
ref.: reference allele, fav.: favourable allele, n.s.: not significant. aFigures in brackets are p values after Bonferroni adjustment
(12 SNPs tested). SNPs included in the final prediction model are printed in bold (see Table 2). All SNPs were analysed using
an additive allelic model on the logit scale.
Supplementary Table 3b: Combined rs12913832/rs1800407 genotype and blue eye colour (stage 2)
rs12913832
(HERC2)
GG
GA
AA
a
rs1800407
(OCA2)
N
Probability
(blue eyes)a
AA
GA
GG
AA
GA
GG
AA
GA
GG
0
6
70
0
3
19
0
0
2
1.00
0.98
0.90
0.84
0.52
0.18
0.12
0.027
0.0055
Prevalence (blue eyes)
observed
expectedb
0
0
5
5.9
61
63.0
0
0
2
1.6
5
3.4
0
0
0
0
0
0.0
Conditional probability of blue eye colour, given the respective genotype, as calculated from the prediction model derived in
stage 1 (see Table 2). bThe expected prevalence equals the product of the genotype prevalence (N) and the probability of blue
eyes.
Supplementary Table 4: Genotype distribution, by eye colour, of the 12 candidate SNPs (all 400
individuals of stage 1 and 2)
SNP-ID
gene
rs12913832
HERC2
rs916977
HERC2
rs1800407
OCA2
rs7495174
OCA2
rs4778138
OCA2
rs4778241
OCA2
rs1805007
MC1R
rs1805008
MC1R
rs1805009
MC1R
rs12203592
IRF4
rs12896399
SLC24A4
rs1393350
TYR
G1
GG
258
0.90
GG
270
0.94
GG
261
0.91
AA
284
0.99
AA
247
0.86
CC
239
0.84
CC
249
0.87
CC
232
0.81
GG
279
0.97
CC
240
0.84
GG
93
0.33
GG
158
0.55
blue (n=287)
G2
G3
A1
GA
AA
G
29
0
545
0.10
0
0.95
GA
AA
G
17
0
557
0.059
0
0.97
GA
AA
G
25
1
547
0.087 0.0035 0.95
AG
GG
A
3
0
571
0.010
0
0.99
AG
GG
A
39
1
533
0.14 0.0035 0.93
CA
AA
C
46
1
524
0.16 0.0035 0.92
CT
TT
C
33
5
531
0.11
0.017 0.93
CT
TT
C
50
5
514
0.17
0.017 0.90
GC
CC
G
8
0
566
0.028
0
0.99
CT
TT
C
45
2
525
0.16 0.0070 0.91
GT
TT
G
134
59
320
0.47
0.21
0.56
GA
AA
G
118
11
434
0.41
0.038 0.76
A2
A
29
0.051
A
17
0.030
A
27
0.047
G
3
0.0052
G
41
0.071
A
48
0.084
T
43
0.075
T
60
0.10
C
8
0.014
T
49
0.085
T
252
0.44
A
140
0.24
G1
GG
26
0.5
GG
39
0.75
GG
47
0.90
AA
45
0.87
AA
37
0.71
CC
32
0.62
CC
42
0.81
CC
45
0.87
GG
51
0.98
CC
46
0.88
GG
20
0.38
GG
31
0.60
green (n=52)
G2
G3
A1
GA
AA
G
26
0
78
0.5
0
0.75
GA
AA
G
13
0
91
0.25
0
0.88
GA
AA
G
5
0
99
0.096
0
0.95
AG
GG
A
7
0
97
0.13
0
0.93
AG
GG
A
15
0
89
0.29
0
0.86
CA
AA
C
19
1
83
0.37 0.019 0.80
CT
TT
C
9
1
93
0.17 0.019 0.89
CT
TT
C
7
0
97
0.13
0
0.93
GC
CC
G
1
0
103
0.019
0
0.99
CT
TT
C
6
0
98
0.12
0
0.94
GT
TT
G
22
10
62
0.42
0.19 0.60
GA
AA
G
18
3
80
0.35 0.058 0.77
13
A2
A
26
0.25
A
13
0.13
A
5
0.048
G
7
0.067
G
15
0.14
A
21
0.20
T
11
0.11
T
7
0.067
C
1
0.0096
T
6
0.058
T
42
0.40
A
24
0.23
G1
GG
5
0.082
GG
27
0.44
GG
52
0.85
AA
39
0.64
AA
28
0.46
CC
27
0.44
CC
51
0.84
CC
45
0.74
GG
61
1
CC
55
0.90
GG
28
0.46
GG
34
0.56
brown (n=61)
G2
G3
A1
GA
AA
G
46
10
56
0.75
0.16 0.46
GA
AA
G
30
4
84
0.49 0.066 0.69
GA
AA
G
9
0
113
0.15
0
0.93
AG
GG
A
22
0
100
0.36
0
0.82
AG
GG
A
30
3
86
0.49 0.049 0.70
CA
AA
C
30
4
84
0.49 0.066 0.69
CT
TT
C
9
1
111
0.15 0.016 0.91
CT
TT
C
14
2
104
0.23 0.033 0.85
GC
CC
G
0
0
122
0
0
1
CT
TT
C
5
1
115
0.082 0.016 0.94
GT
TT
G
25
8
81
0.41
0.13 0.66
GA
AA
G
21
6
89
0.34 0.098 0.73
A2
A
66
0.54
A
38
0.31
A
9
0.074
G
22
0.18
G
36
0.30
A
38
0.31
T
11
0.090
T
18
0.15
C
0
0
T
7
0.057
T
41
0.34
A
33
0.27
Supplementary Table 5: Genotype-phenotype association analysis of red tint in hair colour
Supplementary Table 5a: Association between SNP genotype and red tint (stage 1)
SNP-ID (gene)
rs12913832 (HERC2)
rs916977 (HERC2)
rs1800407 (OCA2)
rs7495174 (OCA2)
rs4778138 (OCA2)
rs4778241 (OCA2)
rs1805007 (MC1R)
rs1805008 (MC1R)
rs1805009 (MC1R)
rs12203592 (IRF4)
rs12896399 (SLC24A4)
rs1393350 (TYR)
Alleles
ref. fav.
C
T
C
T
-
Regression p value
simple
multiple§
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
<1.010-5
6.010-4
n.s.
n.s.
n.s.
n.s.
<1.010-5 (1.210-4)
1.010-4 (0.0012)
n.s.
n.s.
n.s.
n.s.
ref.: reference allele, fav.: favourable allele, n.s.: not significant. §Figures in brackets are p values after Bonferroni adjustment
(12 SNPs tested). SNPs included in the final prediction model are printed in bold (see Table 2). The p value for the interaction
between rs1805007 and rs1805008 equalled 0.020 (0.24). All SNPs were analyzed using an additive allelic model on the logit
scale.
Supplementary Table 5b: Combined rs1805007/rs1805008 genotype and red tint in hair colour
(stage 2)
rs1805007
(MC1R)
CC
CT
TT
rs1805008
(MC1R)
N
Probability
(red tint)$
CC
CT
TT
CC
CT
TT
CC
CT
TT
62
22
4
9
1
0
2
0
0
0.14
0.36
0.67
0.47
0.76
0.92
0.82
0.94
0.98
$
Prevalence (red tint)
observed expected&
10
8.68
9
7.92
3
2.68
6
4.23
1
0.76
0
0
2
1.64
0
0
0
0
Conditional probability of red tint, given the respective genotype, as calculated from the prediction model derived in stage 1
(see Table 2). &The expected prevalence equals the product of the genotype prevalence (N) and the probability of red tint.
14
Supplementary Table 6: Genotype distribution, by red tint, of the 12 candidate SNPs (all 400
individuals of stage 1 and 2)
SNP-ID
gene
rs12913832
HERC2
rs916977
HERC2
rs1800407
OCA2
rs7495174
OCA2
rs4778138
OCA2
rs4778241
OCA2
rs1805007
MC1R
rs1805008
MC1R
rs1805009
MC1R
rs12203592
IRF4
rs12896399
SLC24A4
rs1393350
TYR
G1
GG
76
0.75
GG
88
0.86
GG
92
0.90
AA
96
0.94
AA
81
0.79
CC
77
0.75
CC
68
0.67
CC
68
0.67
GG
97
0.95
CC
85
0.83
GG
37
0.36
GG
57
0.56
red tint (n=102)
G2
G3
A1
GA
AA
G
25
1
177
0.25 0.0098 0.87
GA
AA
G
14
0
190
0.14
0
0.93
GA
AA
G
10
0
194
0.098
0
0.95
AG
GG
A
6
0
198
0.059
0
0.97
AG
GG
A
21
0
183
0.21
0
0.90
CA
AA
C
25
0
179
0.25
0
0.88
CT
TT
C
28
6
164
0.27
0.059 0.80
CT
TT
C
28
6
164
0.27
0.059 0.80
GC
CC
G
5
0
199
0.049
0
0.98
CT
TT
C
15
2
185
0.15
0.020 0.91
GT
TT
G
45
20
119
0.44
0.20
0.58
GA
AA
G
42
3
156
0.41
0.029 0.76
A2
A
27
0.13
A
14
0.069
A
10
0.049
G
6
0.029
G
21
0.10
A
25
0.12
T
40
0.20
T
40
0.20
C
5
0.025
T
19
0.093
T
85
0.42
A
48
0.24
15
G1
GG
213
0.71
GG
248
0.83
GG
268
0.90
AA
272
0.91
AA
231
0.78
CC
221
0.74
CC
274
0.92
CC
254
0.85
GG
294
0.99
CC
256
0.86
GG
104
0.35
GG
166
0.56
no red tint (n=298)
G2
G3
A1
GA
AA
G
76
9
502
0.26
0.030 0.84
GA
AA
G
46
4
542
0.15
0.013 0.91
GA
AA
G
29
1
565
0.097 0.0034 0.95
AG
GG
A
26
0
570
0.087
0
0.96
AG
GG
A
63
4
525
0.21
0.013 0.88
CA
AA
C
70
6
512
0.24
0.020 0.86
CT
TT
C
23
1
571
0.077 0.0034 0.96
CT
TT
C
43
1
551
0.14 0.0034 0.92
GC
CC
G
4
0
592
0.013
0
0.99
CT
TT
C
41
1
553
0.14 0.0034 0.93
GT
TT
G
136
57
344
0.46
0.19
0.58
GA
AA
G
115
17
447
0.39
0.057 0.75
A2
A
94
0.16
A
54
0.091
A
31
0.052
G
26
0.044
G
71
0.12
A
82
0.14
T
25
0.042
T
45
0.076
C
4
0.0067
T
43
0.072
T
250
0.42
A
149
0.25
Supplementary Table 7: Genotype distribution, by hair colour (light-dark component), of the 12
candidate SNPs (all 400 individuals of stage 1 and 2)
SNP-ID
gene
rs12913832
HERC2
rs916977
HERC2
rs1800407
OCA2
rs7495174
OCA2
rs4778138
OCA2
rs4778241
OCA2
rs1805007
MC1R
rs1805008
MC1R
rs1805009
MC1R
rs12203592
IRF4
rs12896399
SLC24A4
rs1393350
TYR
G1
GG
188
0.82
GG
206
0.90
GG
209
0.91
AA
221
0.96
AA
193
0.84
CC
184
0.80
CC
196
0.85
CC
179
0.78
GG
223
0.97
CC
210
0.91
GG
68
0.30
GG
127
0.55
blond(n=230)
G2
G3
A1
GA
AA
G
41
1
417
0.18 0.0043 0.91
GA
AA
G
24
0
436
0.10
0
0.95
GA
AA
G
20
1
438
0.087 0.0043 0.95
AG
GG
A
9
0
451
0.039
0
0.98
AG
GG
A
36
1
422
0.16 0.0043 0.92
CA
AA
C
45
1
413
0.20 0.0043 0.90
CT
TT
C
30
4
422
0.13
0.017 0.92
CT
TT
C
44
7
402
0.19
0.030 0.87
GC
CC
G
7
0
453
0.030
0
0.98
CT
TT
C
20
0
440
0.087
0
0.96
GT
TT
G
107
55
243
0.47
0.24
0.53
GA
AA
G
90
13
344
0.39
0.057 0.75
A2
A
43
0.093
A
24
0.052
A
22
0.048
G
9
0.020
G
38
0.083
A
47
0.10
T
38
0.083
T
58
0.13
C
7
0.015
T
20
0.043
T
217
0.47
A
116
0.25
G1
GG
95
0.64
GG
119
0.80
GG
131
0.89
AA
132
0.89
AA
108
0.73
CC
108
0.73
CC
129
0.87
CC
123
0.83
GG
147
0.99
CC
118
0.80
GG
62
0.42
GG
82
0.55
brown (n=148)
G2
G3
A1
GA
AA
G
44
9
234
0.30
0.061 0.79
GA
AA
G
25
4
263
0.17
0.027 0.89
GA
AA
G
17
0
279
0.11
0
0.94
AG
GG
A
16
0
280
0.11
0
0.95
AG
GG
A
38
2
254
0.26
0.014 0.86
CA
AA
C
34
5
250
0.23
0.034 0.85
CT
TT
C
17
2
275
0.11
0.014 0.93
CT
TT
C
25
0
271
0.17
0
0.92
GC
CC
G
1
0
295
0.0068
0
1.00
CT
TT
C
27
3
263
0.18
0.020 0.89
GT
TT
G
63
22
187
0.43
0.15 0.64
GA
AA
G
60
6
224
0.41
0.041 0.76
Two individuals with pure red hair colour each were removed from stage 1 and 2.
16
A2
A
62
0.21
A
33
0.11
A
17
0.057
G
16
0.054
G
42
0.14
A
44
0.15
T
21
0.071
T
25
0.084
C
1
0.0034
T
33
0.11
T
107
0.36
A
72
0.24
G1
GG
4
0.22
GG
8
0.44
GG
17
0.94
AA
11
0.61
AA
7
0.39
CC
4
0.22
CC
15
0.83
CC
16
0.89
GG
18
1
CC
10
0.56
GG
11
0.61
GG
12
0.67
black (n=18)
G2
G3
A1
GA
AA
G
14
0
22
0.78
0
0.61
GA
AA
G
10
0
26
0.56
0
0.72
GA
AA
G
1
0
35
0.056
0
0.97
AG
GG
A
7
0
29
0.39
0
0.81
AG
GG
A
10
1
24
0.56 0.056 0.67
CA
AA
C
14
0
22
0.78
0
0.61
CT
TT
C
3
0
33
0.17
0
0.92
CT
TT
C
2
0
34
0.11
0
0.94
GC
CC
G
0
0
36
0
0
1
CT
TT
C
8
0
28
0.44
0
0.78
GT
TT
G
7
0
29
0.39
0
0.81
GA
AA
G
5
1
29
0.28 0.056 0.81
A2
A
14
0.39
A
10
0.28
A
1
0.028
G
7
0.19
G
12
0.33
A
14
0.39
T
3
0.083
T
2
0.056
C
0
0
T
8
0.22
T
7
0.19
A
7
0.19
Supplementary Table 8: Genotype-phenotype association analysis of hair colour (light-dark
component)
Supplementary Table 8a: Association between SNP genotype and hair colour (stage 1)
Alleles
SNP-ID (gene)
ref.
fav.
rs12913832 (HERC2)
A
G
rs916977 (HERC2)
rs1800407 (OCA2)
rs7495174 (OCA2)
rs4778138 (OCA2)
rs4778241 (OCA2)
rs1805007 (MC1R)
rs1805008 (MC1R)
rs1805009 (MC1R)
A
G
G
A
C
G
G
A
A
C
T
C
rs12203592 (IRF4)
T
C
rs12896399 (SLC24A4)
G
T
rs1393350 (TYR)
-
-
blond vs. non-blond
simple
multiple§
6.010-4
0.0026
(0.0072)
0.0076
n.s.
n.s.
n.s.
0.032
n.s.
0.031
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
0.0025
0.0032
(0.030)
0.0044
0.0022
(0.053)
n.s.
n.s.
Regression p value
ordinal regression
simple
multiple§
<1.010-5
1.010-5
(1.210-4)
-5
n.s.
4.010
n.s.
n.s.
0.0047
n.s.
-4
n.s.
7.810
0.0071
n.s.
n.s.
0.040 (0.48)
n.s.
0.031 (0.37)
n.s.
0.040 (0.48)
<1.010-5
8.010-5
(1.210-4)
0.0032
0.0094 (0.11)
n.s.
n.s.
linear regression
simple
multiple§
<1.010-5
<1.010-5
(1.210-4)
-5
n.s.
<1.010
n.s.
n.s.
0.0024
n.s.
-4
n.s.
3.010
0.0016
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
-5
<1.010
1.010-4
-4
(1.210 )
0.0050
0.0025
(0.060)
n.s.
n.s.
ref.: reference allele, fav.: favourable allele for light hair colour, n.s.: not significant. §Figures in brackets are p values after
Bonferroni adjustment (12 SNPs tested). SNPs included in the final prediction model are printed in bold (see Table 2). Shown
are the p values for the analysis of blond vs. non-blond, ordinal regression analysis of the nine hair types and linear regression
analysis treating the nine hair types as a quantitative variable. All SNPs were analyzed using an additive allelic model on the
logit scale. Two individuals with pure red hair colour were removed from stage 1.
Supplementary Table 8b: Combined rs12913832/rs12203592 genotype and hair colour (stage 2)
rs12913832
(HERC2)
GG
GA
AA
rs12203592
(IRF4)
N
CC
CT
TT
CC
CT
TT
CC
CT
TT
58
7
1
20
8
1
3
0
0
Prevalence (blond)
n=52
prob.
obs.
exp.
0.69
43
40.02
0.42
3
2.94
0.19
0
0.19
0.50
5
10.00
0.24
1
1.92
0.091
0
0.091
0.29
0
0.87
0.11
0
0
0.038
0
0
Prevalence (brown)
n=36
prob.
obs.
exp.
0.30
14
17.40
0.54
4
3.78
0.73
1
0.73
0.46
12
9.20
0.67
1
5.36
0.75
1
0.75
0.59
3
1.77
0.69
0
0
0.67
0
0
Prevalence (black)
n=10
prob.
obs.
exp.
0.014
1
0.81
0.036
0
0.25
0.073
0
0.073
0.043
3
0.86
0.093
6
0.74
0.16
0
0.16
0.12
0
0.36
0.20
0
0
0.29
0
0
prob.: probability of a particular hair colour, given the respective genotype, as calculated from the prediction model derived in
stage 1 (see Table 2) by multinomial regression, obs.: observed prevalence of the respective hair colour, exp.: expected
prevalence, equal to the product of the genotype prevalence (N) and the probability of the respective hair colour. Two
individuals with pure red hair colour were removed from stage 2.
17
Supplementary Table 9: Genotype distribution, by skin colour, of the 12 candidate SNPs (all 400 individuals of stage 1 and 2)
SNP-ID
gene
rs12913832
HERC2
rs916977
HERC2
rs1800407
OCA2
rs7495174
OCA2
rs4778138
OCA2
rs4778241
OCA2
rs1805007
MC1R
rs1805008
MC1R
rs1805009
MC1R
rs12203592
IRF4
rs12896399
SLC24A4
rs1393350
TYR
G1
GG
17
0.89
GG
17
0.89
GG
16
0.84
AA
19
1
AA
18
0.95
CC
16
0.84
CC
14
0.74
CC
8
0.42
GG
17
0.89
CC
14
0.74
GG
3
0.16
GG
8
0.42
type I (n=19)
G2
G3
A1
GA
AA
G
2
0
36
0.11
0
0.95
GA
AA
G
2
0
36
0.11
0
0.95
GA
AA
G
3
0
35
0.16
0
0.92
AG
GG
A
0
0
38
0
0
1
AG
GG
A
1
0
37
0.053
0
0.97
CA
AA
C
3
0
35
0.16
0
0.92
CT
TT
C
4
1
32
0.21 0.053 0.84
CT
TT
C
10
1
26
0.53 0.053 0.68
GC
CC
G
2
0
36
0.11
0
0.95
CT
TT
C
4
1
32
0.21 0.053 0.84
GT
TT
G
12
4
18
0.63
0.21 0.47
GA
AA
G
10
1
26
0.53 0.053 0.68
A2
A
2
0.053
A
2
0.053
A
3
0.079
G
0
0
G
1
0.026
A
3
0.079
T
6
0.16
T
12
0.32
C
2
0.053
T
6
0.16
T
20
0.53
A
12
0.32
G1
GG
101
0.80
GG
113
0.90
GG
114
0.90
AA
122
0.97
AA
107
0.85
CC
106
0.84
CC
99
0.79
CC
96
0.76
GG
124
0.98
CC
103
0.82
GG
49
0.39
GG
78
0.62
type II (n=126)
G2
G3
A1
GA
AA
G
23
2
225
0.18
0.016 0.89
GA
AA
G
13
0
239
0.10
0
0.95
GA
AA
G
12
0
240
0.095
0
0.95
AG
GG
A
4
0
248
0.032
0
0.98
AG
GG
A
19
0
233
0.15
0
0.92
CA
AA
C
19
1
231
0.15 0.0079 0.92
CT
TT
C
22
5
220
0.17
0.040 0.87
CT
TT
C
25
5
217
0.20
0.040 0.86
GC
CC
G
2
0
250
0.016
0
0.99
CT
TT
C
21
2
227
0.17
0.016 0.90
GT
TT
G
50
26
148
0.40
0.21
0.59
GA
AA
G
42
6
198
0.33
0.048 0.79
A2
A
27
0.11
A
13
0.052
A
12
0.048
G
4
0.016
G
19
0.075
A
21
0.083
T
32
0.13
T
35
0.14
C
2
0.0079
T
25
0.099
T
102
0.41
A
54
0.21
18
G1
GG
168
0.70
GG
198
0.82
GG
217
0.90
AA
216
0.90
AA
184
0.76
CC
171
0.71
CC
216
0.90
CC
205
0.85
GG
236
0.98
CC
214
0.89
GG
80
0.33
GG
131
0.54
type III (n=241)
G2
G3
A1
GA
AA
G
66
7
402
0.27
0.029 0.83
GA
AA
G
39
4
435
0.16
0.017 0.90
GA
AA
G
23
1
457
0.095 0.0041 0.95
AG
GG
A
25
0
457
0.10
0
0.95
AG
GG
A
53
4
421
0.22
0.017 0.87
CA
AA
C
64
5
406
0.27
0.021 0.85
CT
TT
C
24
1
456
0.10 0.0041 0.95
CT
TT
C
35
1
445
0.15 0.0041 0.92
GC
CC
G
5
0
477
0.021
0
0.99
CT
TT
C
27
0
455
0.11
0
0.94
GT
TT
G
116
45
276
0.48
0.19
0.57
GA
AA
G
98
12
360
0.41
0.050 0.75
A2
A
80
0.17
A
47
0.098
A
25
0.052
G
25
0.052
G
61
0.13
A
74
0.15
T
26
0.054
T
37
0.077
C
5
0.010
T
27
0.056
T
206
0.43
A
122
0.25
G1
GG
3
0.21
GG
8
0.57
GG
13
0.93
AA
11
0.79
AA
3
0.21
CC
5
0.36
CC
13
0.93
CC
13
0.93
GG
14
1
CC
10
0.71
GG
9
0.64
GG
6
0.43
type IV (n=14)
G2
G3
A1
GA
AA
G
10
1
16
0.71 0.071 0.57
GA
AA
G
6
0
22
0.43
0
0.79
GA
AA
G
1
0
27
0.071
0
0.96
AG
GG
A
3
0
25
0.21
0
0.89
AG
GG
A
11
0
17
0.79
0
0.61
CA
AA
C
9
0
19
0.64
0
0.68
CT
TT
C
1
0
27
0.071
0
0.96
CT
TT
C
1
0
27
0.071
0
0.96
GC
CC
G
0
0
28
0
0
1
CT
TT
C
4
0
24
0.29
0
0.86
GT
TT
G
3
2
21
0.21
0.14 0.75
GA
AA
G
7
1
19
0.50 0.071 0.68
A2
A
12
0.43
A
6
0.21
A
1
0.036
G
3
0.11
G
11
0.39
A
9
0.32
T
1
0.036
T
1
0.036
C
0
0
T
4
0.14
T
7
0.25
A
9
0.32
Supplementary Table 10: Genotype-phenotype association analysis of skin colour
Supplementary Table 10a: Association between SNP genotype and skin colour (stage 1)
Alleles
SNP-ID (gene)
rs12913832 (HERC2)
rs916977 (HERC2)
rs1800407 (OCA2)
rs7495174 (OCA2)
rs4778138 (OCA2)
rs4778241 (OCA2)
rs1805007 (MC1R)
rs1805008 (MC1R)
rs1805009 (MC1R)
rs12203592 (IRF4)
rs12896399 (SLC24A4)
rs1393350 (TYR)
ref.
A
A
G
G
A
C
C
C
-
fav.
G
G
A
A
C
T
T
T
-
Regression p value
types I-II vs. types III-IV
ordinal regression
simple
multiple§
simple
multiple§
n. s.
n. s.
0.0046
n. s.
n. s.
n. s.
0.039
n. s.
n. s.
n. s.
n. s.
n. s.
0.031
0.0067 (0.080)
0.010
n. s.
0.024
n. s.
2.810-4 4.910-4 (0.0059)
0.044
n. s.
0.0060
n. s.
0.0041
0.0023 (0.028)
0.0060
0.0035 (0.042)
6.010-4 1.010-4 (0.0012) 1.210-4 6.010-5 (7.210-4)
n. s.
n. s.
n. s.
n. s.
0.034
0.027 (0.32)
n. s.
n. s.
n. s.
n. s.
n. s.
n. s.
n. s.
n. s.
n. s.
n. s.
ref.: reference allele, fav.: favourable allele for fair skin type, n.s.: not significant. §Figures in brackets are p values after
Bonferroni adjustment (12 SNPs tested). SNPs included in the final prediction model are printed in bold (see Table 2). All
SNPs were analyzed using an additive allelic model on the logit scale.
Supplementary Table 10b: Combined rs1805007/rs1805008/rs4778138 genotype and skin colour
(stage 2)
rs1805007
(MC1R)
rs1805008
(MC1R)
CC
CC
CT
TT
CC
CT
CT
TT
CC
TT
CT
TT
rs4778138
(OCA2)
AA
AG
GG
AA
AG
GG
AA
AG
GG
AA
AG
GG
AA
AG
GG
AA
AG
GG
AA
AG
GG
AA
AG
GG
AA
AG
GG
N
48
10
3
13
8
0
2
0
0
9
4
0
1
0
0
0
0
0
2
0
0
0
0
0
0
0
0
19
Probability
(types I-II)$
0.25
0.14
0.074
0.48
0.31
0.18
0.73
0.56
0.39
0.46
0.29
0.17
0.71
0.54
0.36
0.87
0.77
0.62
0.69
0.52
0.34
0.86
0.75
0.60
0.95
0.90
0.81
Prevalence (types I-II)
observed expected&
26
12.00
3
1.40
0
0.22
8
6.24
3
2.48
0
0
2
1.46
0
0
0
0
7
4.14
2
1.16
0
0
1
0.71
0
0
0
0
0
0
0
0
0
0
2
1.38
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
$
Conditional probability of types I-II, given the respective genotype, as calculated from the prediction model derived in stage 1
(see Table 2). &The expected prevalence equals the product of the genotype prevalence (N) and the probability of types I-II.
Supplementary Table 11: Composition of the four phenotype clusters identified by
multidimensional scaling analysis (all 400 individuals of stage 1 and 2)
eye colour
blue
green
brown
hair colour
blond
brown
black
skin type
type I
type II
type III
type IV
blue eye colour
red tint
blue eye colour
no red tint
no blue eye colour
red tint
no blue eye colour
no red tint
71 (100%)
0 (0%)
0 (0%)
214 (100%)
0 (0%)
0 (0%)
0 (0%)
14 (52%)
13 (48%)
0 (0%)
46 (55%)
38 (45%)
46 (65%)
23 (32%)
2 (3%)
145 (68%)
64 (30%)
5 (2%)
11 (41%)
16 (59%)
0 (0%)
28 (33%)
45 (54%)
11 (13%)
8 (11%)
35 (49%)
27 (38%)
1 (1%)
6 (3%)
69 (32%)
135 (63%)
4 (2%)
2 (7%)
13 (48%)
10 (37%)
2 (7%)
1 (1%)
9 (11%)
67 (80%)
7 (8%)
Results were obtained from all 400 individuals from stages 1 and 2, except four individuals with pure red hair colour who were
removed from the analysis.
Supplementary Table 12: Association between SNP genotype and blue eye colour (all 400
individuals of stage 1 and 2)
SNP-ID (gene)
rs12913832 (HERC2)
rs916977(HERC2)
rs1800407(OCA2)
rs7495174(OCA2)
rs4778138(OCA2)
rs4778241(OCA2)
rs1805007(MC1R)
rs1805008(MC1R)
rs1805009 (MC1R)
rs12203592(IRF4)
rs12896399 (SLC24A4)
rs1393350 (TYR)
Alleles
ref.
fav.
A
G
A
G
G
A
G
A
G
A
A
C
C
T
-
Regression p value
simple
multiplea
<1.010-5
<1.010-5
n.s.
<1.010-5
<1.010-5
<1.010-5
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
<1.010-5 (1.210-4)
n.s.
0.0072 (0.086)
0.0047 (0.056)
n.s.
n.s.
n.s.
n.s.
n.s.
0.030 (0.36)
n.s.
n.s.
ref.: reference allele, fav.: favourable allele, n.s.: not significant. aFigures in brackets are p values after Bonferroni adjustment
(12 SNPs tested). All SNPs were analysed using an additive allelic model on the logit scale. SNPs included in the final
prediction model are printed in bold (see Table S17).
20
Supplementary Table 13: Association between SNP genotype and red tint (all 400 individuals of
stage 1 and 2)
SNP-ID (gene)
rs12913832 (HERC2)
rs916977 (HERC2)
rs1800407 (OCA2)
rs7495174 (OCA2)
rs4778138 (OCA2)
rs4778241 (OCA2)
rs1805007 (MC1R)
rs1805008 (MC1R)
rs1805009 (MC1R)
rs12203592 (IRF4)
rs12896399 (SLC24A4)
rs1393350 (TYR)
Alleles
ref. fav.
C
T
C
T
-
Regression p value
simple
multiple§
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
<1.010-5
<1.010-5
n.s.
n.s.
n.s.
n.s.
<1.010-5 (1.210-4)
<1.010-5 (1.210-4)
n.s.
n.s.
n.s.
n.s.
ref.: reference allele, fav.: favourable allele, n.s.: not significant. §Figures in brackets are p values after Bonferroni adjustment
(12 SNPs tested). The p value for the interaction between rs1805007 and rs1805008 equalled 0.023 (0.28). All SNPs were
analyzed using an additive allelic model on the logit scale. SNPs included in the final prediction model are printed in bold (see
Table S17).
Supplementary Table 14: Association between SNP genotype and the light-dark component of hair
colour (all 400 individuals of stage 1 and 2)
SNP-ID (gene)
Alleles
blond vs. non-blond
simple
multiple§
<1.010-5
<1.010-5
(1.210-4)
-4
n.s.
1.010
n.s.
n.s.
n.s.
6.010-4
-4
n.s.
5.010
0.0033
n.s.
n.s.
n.s.
ref.
fav.
rs12913832 (HERC2)
A
G
rs916977 (HERC2)
rs1800407 (OCA2)
rs7495174 (OCA2)
rs4778138 (OCA2)
rs4778241 (OCA2)
rs1805007 (MC1R)
A
G
G
A
-
G
A
A
C
-
rs1805008 (MC1R)
C
T
n.s.
n.s.
rs1805009 (MC1R)
G
C
n.s.
rs12203592 (IRF4)
T
C
<1.010-5
rs12896399 (SLC24A4)
G
T
6.010-4
rs1393350 (TYR)
-
-
n.s.
n.s.
<1.010-5
(1.210-4)
0.0030
(0.036)
n.s.
Regression p value
ordinal regression
simple
multiple§
<1.010-5
<1.010-5
(1.210-4)
-5
n.s.
<1.010
n.s.
n.s.
<1.010-5 0.035 (0.42)
n.s.
<1.010-5
-5
n.s.
2.010
n.s.
n.s.
3.010-5
1.010-4
(3.610-4)
0.034
0.037 (0.44)
<1.010-5
<1.010-5
(1.210-4)
0.0041
5.010-4
(0.49)
n.s.
n.s.
linear regression
simple
multiple§
<1.010-5
<1.010-5
(1.210-4)
-5
n.s.
<1.010
n.s.
n.s.
n.s.
<1.010-5
-5
0.024 (0.29)
<1.010
-5
n.s.
<1.010
n.s.
n.s.
2.010-4
4.010-4
(0.0024)
0.028
0.049 (0.59)
<1.010-5
<1.010-5
(1.210-4)
0.0032
2.010-4
(0.038)
n.s.
n.s.
ref.: reference allele, fav.: favourable allele for light hair colour, n.s.: not significant. §Figures in brackets are p values after
Bonferroni adjustment (12 SNPs tested). Shown are the p values for the analysis of blond vs. non-blond, ordinal regression
analysis of the nine hair types and linear regression analysis treating the nine hair types as a quantitative variable. All SNPs
were analyzed using an additive allelic model on the logit scale. Two individuals with pure red hair colour each were removed
from stage 1 and 2. SNPs included in the final prediction model are printed in bold (see Table S17).
21
Supplementary Table 15: Association between SNP genotype and skin colour (all 400 individuals of
stage 1 and 2)
SNP-ID (gene)
Alleles
rs12913832 (HERC2)
rs916977 (HERC2)
rs1800407 (OCA2)
rs7495174 (OCA2)
rs4778138 (OCA2)
rs4778241 (OCA2)
rs1805007 (MC1R)
ref.
A
A
G
G
A
C
fav.
G
G
A
A
C
T
rs1805008 (MC1R)
C
T
rs1805009 (MC1R)
rs12203592 (IRF4)
rs12896399 (SLC24A4)
rs1393350 (TYR)
C
-
T
-
Regression p value
types I-II vs. types III-IV
ordinal regression
§
simple
multiple
simple
multiple§
0.0022
n. s.
0.0062 (0.074)
5.010-5
0.014
n. s.
0.0031
n. s.
n. s.
n. s.
n. s.
n. s.
-4
-4
0.0034 2.010 (0.0024) 7.810
n. s.
0.0016
n. s.
0.0013 (0.016)
<1.010-5
n. s.
n. s.
8.010-4
1.910-4
-4
-4
-4
6.010
1.010 (0.0012) 2.410
9.010-5 (0.0011)
-5
<1.010
2.010-4
1.010-5 1.010-5 (1.210-4)
(1.210-4)
n. s.
n. s.
n. s.
n. s.
0.027
0.014 (0.17)
0.040
0.024 (0.29)
n. s.
n. s.
n. s.
n. s.
n. s.
n. s.
n. s.
n. s.
ref.: reference allele, fav.: favourable allele for fair skin type, n.s.: not significant. §Figures in brackets are p values after
Bonferroni adjustment (12 SNPs tested). All SNPs were analyzed using an additive allelic model on the logit scale. SNPs
included in the final prediction model are printed in bold (see Table S17).
Supplementary Table 16: Associations between dichotomous pigmentation traits (all 400 individuals
of stage 1 and 2)
Pigmentation trait
p (padj)
eye colour (blue)
p (padj)a
hair colour (blond)
<1.010-5 (3.010-5)
hair colour (red tint)
8.010-4 (0.0024)
skin type (fair, I-II)
<1.010-5 (3.010-5)
hair colour (blond)b
4.010-4 (1.210-3)
eye colour (blue)
skin type (fair, I-II)
<1.010-5 (3.010-5)
0.0066 (0.020)
skin type (fair, I-II)d
0.00527 (0.016)
0.0025 (0.0075)c
eye colour (blue)
hair colour (red)
hair colour (blond)
<1.010-5 (3.010-5)
0.00417 (0.013)
9.410-4 (0.0028)c
0.00169 (0.0051)d
2.510-4 (7.510-4)d
0.015 (0.045)d
padj: Bonferroni-adjusted p value (three tests). All associations were evaluated in a multiple logistic regression analysis treating
the remaining three traits as influential variables, and by backward selection. aP values and ORs are from a multiple logistic
regression analysis including the selected SNPs from Tables S12-S15 (cf. also Table S17) as influential variables. bIn the
analysis of the light-dark component of hair colour, two individuals with pure red hair colour each were removed from stage 1
and 2. cMultiple logistic regression analysis eventually disregarded rs1805008 (MC1R) after backward selection. dMultiple
logistic regression analysis eventually disregarded rs7495174 (OCA2) and rs4778138 (OCA2) after backward selection.
22
Supplementary Table 17: Capability of selected models to predict dichotomous pigmentation
phenotypes (all 400 individuals of stage 1 and 2)
Phenotype
Prev.
rs12913832 (HERC2)
eye colour
(blue)
72
hair colour
(red tint)
26
hair colour
(blond)
skin type
(fair, I-II)
Predictors
58
36
rs12913832 (HERC2)
hair colour (blond)
rs1805007 (MC1R)
rs1805008 (MC1R)
rs1805007 (MC1R)
rs1805008 (MC1R)
skin type (fair, I-II)
rs12913832 (HERC2)
rs1805008 (MC1R)
rs12203592 (IRF4)
rs12896399 (SLC24A4)
rs12913832 (HERC2)
rs12203592 (IRF4)
rs12896399 (SLC24A4)
eye colour ((blue)
skin type (fair, I-II)
rs1805008 (MC1R)
rs1805007 (MC1R)
rs7495174 (OCA2)
rs4778138 (OCA2)
rs1805008 (MC1R)
rs1805007 (MC1R)
eye colour (blue)
hair colour (red)
hair colour (blond)
median(range)
all (95% CI)
median(range)
all (95% CI)
median(range)
all (95% CI)
median(range)
all (95% CI)
Sensitivity
91 (80-93)
90 (86-93)
91 (80-93)
90 (86-93)
23 (9-88)
39 (29-49)
27 (0-63)
28 (21-37)
Specificity
80 (29-91)
90 (86-93)
80 (29-91)
72 (64-80)
97 (83-100)
92 (89-95)
93 (81-100)
95 (93-97)
Accuracy
86 (78-93)
85 (82-89)
86 (78-93)
85 (82-88)
73 (70-95)
78 (75-82)
75 (73-85)
76 (71-82)
AUC
84 (60-92)
82 (77-86)
84 (60-92)
85 (81-90)
71 (52-90)
72 (67-78)
76 (60-88)
77 (71-82)
median(range)
all (95% CI)
78 (71-91)
82 (77-87)
47 (21-65)
47 (39-54)
68 (55-75)
67 (63-71)
69 (54-82)
70 (65-76)
median(range)
all (95% CI)
79 (30-94)
80 (74-85)
60 (33-69)
54 (47-61)
69 (40-75)
69 (65-73)
73 (44-88)
74 (70-79)
median(range)
all (95% CI)
86 (55-93)
84 (80-89)
45 (19-61)
42 (34-50)
71 (53-85)
69 (65-73)
66 (47-84)
67 (62-73)
median(range)
all (95% CI)
88 (72-100)
87 (83-91)
44 (22-67)
48 (39-55)
73 (58-80)
73 (69-77)
68 (59-88)
73 (68-78)
Prev.: prevalence, AUC: area under the receiver operating characteristic curve, median (range): median and range of ten cross
validation samples, all (95% CI): values for all 400 individuals with 95% confidence interval in brackets. Two individuals with
pure red hair colour each were removed from stage 1 and 2 for the analysis of blond hair colour. All numerical figures are
percentages.
23