Table 1.
Skin, hair, and eye pigmentation by sex and country.
Figure 1.
Distribution of skin, hair, and eye pigmentation.
Skin pigmentation histogram (A) and boxplot by country of sampling and sex (B) in 469 individuals showing the normality of the phenotype distribution and the differences between sexes and among countries. Males (M) have consistently lighter pigmentation (lower scored) than females (F) in all four countries. Among countries, the largest pigmentation difference is with Ireland, where, in our sample, individuals have lighter pigmentation or lower M index on average than in Poland, Italy, or Portugal. Hair pigmentation histogram (C) and boxplot by country (D) in 341 individuals showing the distribution of hair pigmentation and the differences among countries. In our sample, individuals from Northern European countries (Ireland, Poland) have on average lighter hair pigmentation than individuals from Southern European countries (Italy, Portugal). The distributions in males are similar to those in females in all countries except Ireland, where, in our sample, males have darker hair color than females (not shown). Eye pigmentation histogram (E) and boxplot by country (F) in 468 individuals showing the bimodal distribution of eye pigmentation and the differences among countries. Comparison with self-reported phenotypes shows that the two modes of the distribution correspond to blue and brown eye color, while individuals reporting green and hazel eye color have intermediate C’ values. As with hair pigmentation, in our sample, individuals from Northern European countries have on average lighter eye pigmentation than individuals from Southern European countries.
Figure 2.
Comparison of measured and self-reported hair and eye color.
(A) The boxplots of hair pigmentation (M index) binned by self-reported hair color show that there is a good correspondence between categorical hair color and the M index. On average, individuals with red hair color (n = 6) have lower M indices than blond individuals (n = 58). Individuals reporting brown (n = 252) or black (n = 11) hair color have the highest M indices. (B) Similarly for eye pigmentation, the C’ score corresponds well to the self-reported eye color. Individuals reporting blue (n = 123), green (n = 97), hazel (n = 14), and brown (n = 212) eye color have increasing average C’ scores. We note that one individual has a discrepant self-reported color vs. C’ score. This individual self-reported light brown eye color but the C’ score is indicative of most likely blue eye color. This individual was part of the replication cohort and removing the sample from the association study produces similar p-values, only slightly less significant for the OCA2/HERC2 SNPs. Each box in the plot is drawn with a width proportional to the square root of the sample size.
Figure 3.
Population structure of the GWAS samples.
(A) Plots of GWAS individuals on genetic PC1 and PC2 show that individuals largely cluster by country of sampling. PC1 divides the samples according to a North/South geographical axis, while PC2 divides the samples along an East/West geographical axis. Individuals from Ireland, Poland, Italy, and Portugal are colored in yellow, red, blue, and gray, respectively. (B) The plot of individuals on PC1 and PC3 shows that individuals from Portugal tend to have lower values on PC3 than individuals from Italy and Ireland, while individuals from Poland have intermediate values on PC3.
Figure 4.
Percent phenotypic variance explained by increasing numbers of genetic PCs or country of sampling.
Plotted is the percent phenotypic variance explained by genetic PC1 (PC1); PC1 and PC2 (PC1–2); PC1, PC2, and PC3 (PC1–3); PC1, PC2, PC3, and PC4 (PC1–4); or country of sampling (country) in the GWAS cohort. The first three genetic PCs, similar to the country of sampling, explain a remarkably large proportion of the variation in pigmentation: 38.6/37.8%, 24.8/24.1%, 34.5/35.5% for the skin, hair, and eye pigmentation phenotypes by country of sampling/3 PCs. Including PC4 does not explain significantly more phenotypic variation.
Figure 5.
Manhattan plots for the GWAS results for the skin (A), hair (B), and eye (C) pigmentation. The log-transformed p-values from the test of association are plotted as a function of the chromosomal position. Genome-wide significance is defined as the Bonferroni corrected 5% significance threshold (p-value<1.6×10−7) and is indicated as a red line. For skin pigmentation, one SNP on chromosome 3 in the FLNB gene almost reaches genome-wide significance (p-value = 1.8×10−7). No SNP achieves genome-wide significance in the hair pigmentation GWAS. For eye pigmentation, two SNPs, one near the pigmentation gene OCA2 on chromosome 15 and one in the SCIN gene on chromosome 7 achieve genome-wide significance.
Table 2.
GWAS, replication, and combined association results for all signals with p-value<10−5 in the GWAS.