Genetic Architecture of Skin and Eye Color in an African-European Admixed Population

May combine two claims.

Four skin color loci (5p13.3, 11q14.3, 15q13.1, and 15q21.1) and two eye color loci (15q13.1 and 15q21.1) met genome-wide significance P<5.7×10−8 in genotype-based association.

“For genotype-based association, four skin color loci (5p13.3, 11q14.3, 15q13.1, and 15q21.1) and two eye color loci (15q13.1 and 15q21.1) met genome-wide criteria for significance, P<5.7×10−8”

May combine two claims.

The most strongly associated SNPs at 5p13.3 and 15q21.1 represent missense alterations SLC45A2 L374F and SLC24A5 A111T.

“the most strongly associated SNPs at 5p13.3 and 15q21.1 represent missense alterations in two well-known pigmentary genes, SLC45A2 L374F (Figure 4a) and SLC24A5 A111T (Figure 4c)”

May combine two claims.

GRM5 and APBA2 lie 396 kb upstream and 1025 kb upstream of TYR and OCA2, respectively.

“GRM5 and APBA2 lie 396 kb upstream and 1025 kb upstream of TYR and OCA2, respectively”

The quote only partly states this claim.

The pattern of African genomic ancestry in Cape Verde is less skewed than in African Americans from FBPP, where African ancestry ranges 40.6% to 99.3% with median 89.5%.

“in a randomly sampled African-American cohort from the Family Blood Pressure Study (FBPP) [21], African genomic ancestry ranges from 40.6% to 99.3%, with a median of 89.5%”

• cites [21] H Tang, 2006, Racial admixture and its impact on BMI and blood pressure in African and Mexican Americans, Hum Genet sourceNot checked yet

  Not checked in the free run.

• AssumptionRests on an assumption: HapMap CEU and YRI individuals adequately represent the European and African ancestral populations of Cape Verde.

May combine two claims.

Across Cape Verde islands, distributions of African genomic ancestry overlap in range but vary in median from 50.5% in Fogo to 74.4% in Santiago.

“Across different islands, the distributions of African genomic ancestry exhibit substantial overlap in range but vary in their median values, from 50.5% in Fogo to 74.4% in the capital island of Santiago”

• AssumptionRests on an assumption: HapMap CEU and YRI individuals adequately represent the European and African ancestral populations of Cape Verde.

May combine two claims.

The 5p13.3 and 11q14.3 regions affect only skin color, whereas chromosome 15 regions 15q13.1 and 15q21.1 affect both skin and eye color.

“The 5p13.3 and 11q14.3 regions affect only skin color (Figure 4a, 4b), whereas the chromosome 15 regions (15q13.1 and 15q21.1, Figure 4c, 4d) affect both skin and eye color.”

May combine two claims.

Ancestry-based association yields broader peaks than genotype-based association but uncovers the identical set of six loci.

“Ancestry-based association yields broader peaks than genotype-based association, but the distribution of significant regions is strikingly consistent for the two approaches, uncovering the identical set of six loci”

• AssumptionRests on an assumption: HapMap CEU and YRI individuals adequately represent the European and African ancestral populations of Cape Verde.

May combine two claims.

In analyses considering TYR and GRM5 SNPs as covariates, both remain significant (P=3.7×10−6 for rs10831496; P=0.00012 for rs1042602).

“in analyses that consider TYR and GRM5 SNPs as covariates, both remain significant (P = 3.7×10−6 for rs10831496; P = 0.00012 for rs1042602).”

May combine two claims.

Each derived allele at HERC2 (OCA2) and APBA2 (OCA2) is present on a single major extended haplotype (>160 kb for HERC2 and >90 kb for APBA2), whereas ancestral alleles are found on multiple shorter haplotypes.

“Each derived allele is present on a single major extended haplotype (>160 kb in length for HERC2 (OCA2) and >90 kb in length for APBA2 [OCA2]), whereas the ancestral alleles are found on multiple, shorter, haplotypes.”

May combine two claims.

In conditional regression analyses, HERC2 does not affect skin color if eye color is included as a covariate (HERC2 P=0.48, APBA2 P=0.0009); conversely, APBA2 does not affect eye color if skin color is included (APBA2 P=0.083, HERC2 P<2×10−16).

“in conditional regression analyses where genotypes at both loci are included as covariates, HERC2 does not affect skin color if eye color is also included as a covariate (HERC2 P = 0.48, APBA2 P = 0.0009); conversely, APBA2 does not affect eye color if skin color is also included as a covariate (APBA2 P = 0.083, HERC2 P<2×10−16).”

May combine two claims.

For eye color, the ancestral HERC2 (OCA2) allele is mostly dominant over the derived allele, consistent with the near recessive mode of inheritance of blue eye phenotype in Europeans.

“For eye color, the ancestral HERC2 (OCA2) allele is mostly dominant over the derived allele, consistent with the near recessive mode of inheritance of blue eye phenotype in Europeans”

May combine two claims.

The four skin color loci act in an additive fashion; there is no evidence of dominance at any of the loci, nor evidence of opposite-sign epistasis between loci.

“The four skin color loci we identified by association analysis act in an additive fashion: we found no evidence of dominance at any of the loci, nor evidence of opposite-sign epistasis between loci.”

May combine two claims.

The SLC45A2 and APBA2 (OCA2) GWAS signals coincide with sharp CMS signals; broader CMS signals are apparent at SLC24A5 and HERC2 (OCA2).

“The SLC45A2 and APBA2 (OCA2) GWAS signals coincide with sharp CMS signals (Figure 4a, 4d); broader CMS signals are also apparent at SLC24A5 and HERC2 (OCA2) (Figure 4c, 4d).”

May combine two claims.

The derived APBA2 (OCA2) allele is present at low frequencies in most populations of African ancestry and at high frequencies in most populations of Asian and European ancestry.

“The derived APBA2 (OCA2) allele is present at low frequencies in most populations of African ancestry, and at high frequencies in most populations of Asian and European ancestry.”

• AssumptionRests on an assumption: HapMap CEU and YRI individuals adequately represent the European and African ancestral populations of Cape Verde.

May combine two claims.

Measures of linkage disequilibrium between HERC2 and APBA2 loci are not significant (R^2=0.036, D′=0.071 with Χ2=0.517), and extended haplotypes for the derived alleles are separated by ∼800 kb.

“measures of linkage disequilibrium between the two loci are not significant (R∧2 = 0.036, D′ = 0.071 with Χ2 = 0.517), extended haplotypes for the derived alleles are separated by ∼800 kb”

May combine two claims.

In the subset of 592 Cape Verdeans with T-index >0.15, both HERC2 rs12913832 (P=5.23×10−16) and SLC24A5 rs2470102 (P=1.12×10−10) remain highly significant.

“In the subset of 592 Cape Verdeans whose T-index >0.15 (Figure 1, Figure 2), both loci remain highly significant (HERC2 rs12913832, P = 5.23×10−16; and SLC24A5 rs2470102, P = 1.12×10−10)”

May combine two claims.

The effects of SLC24A5 on eye color are semi-dominant, and no interaction was found between this gene and HERC2.

“By contrast, the effects of SLC24A5 on eye color are semi-dominant, and no interaction was found between this gene and HERC2”

May combine two claims.

The proportions of phenotypic variance attributed solely to genotype at each of the four major skin color loci are about 2% each for GRM5-TYR, APBA2 (OCA2), and SLC45A2, and about 7% for SLC24A5.

“the proportions of phenotypic variance attributed solely to genotype at each of the four major skin color loci are quite small, about 2% each for GRM5-TYR, APBA2 (OCA2), and SLC45A2, and about 7% for SLC24A5.”

May combine two claims.

The derived HERC2 (OCA2) allele is absent from African and East Asian populations and appears at high frequency only in Western and Northern Europe.

“By contrast, the derived HERC2 (OCA2) allele is absent from African and East Asian populations, and appears at high frequency only in Western and Northern Europe”

• AssumptionRests on an assumption: HapMap CEU and YRI individuals adequately represent the European and African ancestral populations of Cape Verde.

May combine two claims.

Individual ancestry contributes 44% to skin color variation, and about half of that contribution (22% of total) could also be accounted for by genotype at the four major loci.

“even though individual ancestry contributes 44% to skin color variation, about half of that contribution (22% of the total) could also be accounted for by genotype at the four major loci.”

May combine two claims.

The four major loci contribute a total of 35% to skin color variation, but only 13% above that which could also be accounted for by individual ancestry.

“the four major loci contribute a total of 35% to skin color variation, but only 13% above that which could also be accounted for by individual ancestry.”

May combine two claims.

The GWAS significance peaks for eye color and skin color at 15q13.1 appear distinct, and linkage disequilibrium between the two peaks is weak (R^2=0.112).

“The GWAS significance peaks for eye color and skin color at 15q13.1 appear distinct (Figure 4d), and linkage disequilibrium between the two significance peaks is weak (R∧2 = 0.112 in our sample).”

May combine two claims.

Even though SLC24A5 and SLC45A2 lie on different chromosomes, their genotypes in Cape Verde are correlated with genomic ancestry and with each other.

“even though SLC24A5 and SLC45A2 lie on different chromosomes, their genotypes in the Cape Verde population are correlated with genomic ancestry and, therefore, with each other.”

Classified as probably not a substantial claim. May combine two claims.

The final dataset consists of 685 individuals with skin reflectance measurements, of which 625 have high-quality digital eye photographs.

“The final dataset consists of 685 individuals with skin reflectance measurements, of which 625 have high-quality digital eye photographs.”

May combine two claims.

The 15q13.1 skin color locus has a haplotype set, linkage disequilibrium architecture, and signature of selection that is clearly distinct from the HERC2 (OCA2) eye color locus.

“and has a haplotype set, linkage disequilibrium architecture, and signature of selection that is clearly distinct from the HERC2 (OCA2) eye color locus.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

After applying all quality control filters, 903,837 autosomal SNPs were available for analysis, of which 24,478 had minor allele frequency (MAF) <0.01.

“After applying all quality control filters, 903,837 autosomal SNPs were available for analysis, of which 24,478 had minor allele frequency (MAF) <0.01.”

May combine two claims.

In the Cape Verde sample, we used dense genotype data (>900,000 SNPs) to correct for genomic ancestry, and only the first three loci exhibited a significant association with skin color.

“In the Cape Verde sample, we used dense genotype data (>900,000 SNPs) to correct for genomic ancestry, and only the first three loci exhibited a significant association with skin color.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Our results do not exclude the possibility that variation in KITLG affects skin color.

“Our results do not exclude the possibility that variation in KITLG affects skin color”

Probably not a substantial claim: may not pass S3 (load-bearing).

Effect sizes shown in Figure 5 and Figure 7 for APBA2 (0.155), GRM5-TYR (0.193), SLC24A5 (0.334), and SLC45A2 (0.166) are from a linear regression of the Cape Verde skin color data in which genomic ancestry and genotypes at the four major loci are included as covariates.

“Effect sizes shown in Figure 5 and Figure 7 for APBA2 (0.155) GRM5-TYR (0.193), SLC24A5 (0.334), and SLC45A2 (0.166) are from a linear regression of the Cape Verde skin color data in which genomic ancestry and genotypes at the four major loci are included as covariates.”

May combine two claims.

We carried out an association scan conditioning at all index SNPs, and found no evidence for secondary signals except in the GRM5-TYR region (rs10831496 and rs1042602).

“To evaluate potential secondary signals, we also carried out an association scan conditioning at all index SNPs, and found no evidence for secondary signals except in the GRM5-TYR region (rs10831496 and rs1042602, respectively) as described in the conditional analysis section of the Results.”

May combine two claims.

MM in individuals of European ancestry is smaller and more narrowly distributed, with mean 5.38 and SD 0.27.

“By comparison, MM in individuals of European ancestry is smaller and much more narrowly distributed, with a mean of 5.38 and standard deviation of 0.27 [23].”

• cites [23] SI Candille, 2012, Genome-wide association studies of quantitatively measured skin, eye, and hair pigmentation in four European populations, PLoS ONE sourceNot checked yet

  Not checked in the free run.

May combine two claims.

The direction of effect of the six loci is consistent across all island groups, with the derived allele associated with lighter skin color or a lower T-index.

“the direction of effect is consistent across all groups, with the derived allele associated with lighter skin color or a lower T-index”

May combine two claims.

After removing SNPs surrounding the six loci, genomic control parameters were 1.02 for skin color and 1.00 for eye color.

“After removing SNPs surrounding these six loci, the genomic control parameters were 1.02 and 1.00 for skin color and eye color, respectively.”

May combine two claims.

GWAS results are based on linear regression adjusting for the first three principal components, and adjusting for up to 10 PCs does not qualitatively alter results.

“The genome-wide association results described below are based on a linear regression adjusting for the first three principal components (PC) but the same model adjusting for up to 10 PCs does not qualitatively alter our results.”

• AssumptionRests on an assumption: HapMap CEU and YRI individuals adequately represent the European and African ancestral populations of Cape Verde.

May combine two claims.

The 15q21.1 skin color signal corresponds to SLC24A5 and the 15q13.1 eye color signal corresponds to HERC2 (OCA2).

“As described below, these signals correspond to SLC24A5 for skin color and HERC2 (OCA2) for eye color”

Classified as probably not a substantial claim.

SLC45A2 and SLC24A5 are thought to encode membrane transport proteins that promote melanogenesis.

“thought to encode membrane transport proteins that promote melanogenesis [13], [25]–[27].”

• cites [13], [25]–[27]Not checked yet

  Not checked in the free run.

May combine two claims.

Expression of SLC45A2 and SLC24A5 is pigment cell-specific, and null mutations dramatically impair melanin synthesis in model organisms.

“expression of both genes is pigment cell-specific, and null mutations dramatically impair melanin synthesis in model organisms [13], [28], [29].”

• cites [13], [28], [29]Not checked yet

  Not checked in the free run.

May combine two claims.

Cape Verde has extensive West African/European admixture that has given rise to a broad range in trait values and genomic ancestry proportions.

“We study pigmentary variation in 699 individuals from Cape Verde, where extensive West African/European admixture has given rise to a broad range in trait values and genomic ancestry proportions.”

• Also stated in: § Introduction · plain

  “In this population, extensive genetic admixture between the African and European ancestral populations during the last several centuries [16], [17] has facilitated assortment of pigmentary alleles.”

• cites [16], [17]Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

The HERC2 region has been well recognized to account for Mendelian-like inheritance of blue vs. brown eye color in populations of European ancestry.

“has been well recognized to account for Mendelian-like inheritance of blue vs. brown eye color in populations of European ancestry [7], [8], [10].”

• cites [7], [8], [10]Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

TYR and OCA2 are well-known pigmentary genes for which complete loss-of-function causes albinism in humans and other animals.

“two well-known pigmentary genes for which a complete loss-of-function causes albinism in humans and in other animals [30]–[32].”

• cites [30]–[32]Not checked yet

  Not checked in the free run.

May combine two claims.

In any particular individual, genomic ancestry has predictive value for genotype at each of the four major skin color loci, and vice versa.

“Thus, in any particular individual, genomic ancestry has predictive value for genotype at each of the four major skin color loci, and vice versa.”

May combine two claims.

If skin color in Cape Verde is 80% heritable, the four major loci and individual ancestry explain most of that variation, with only a small amount of missing heritability.

“If skin color in Cape Verde is 80% heritable, the four major loci and individual ancestry explain most of that variation, with only a small amount of “missing heritability””

• AssumptionThe authors state this as an assumption: Skin color in Cape Verde is 80% heritable.

Classified as probably not a substantial claim. May combine two claims.

The failure to replicate previously reported effects could be due to a paucity of allelic variation in Cape Verde ancestral populations (e.g. MC1R), incomplete correction for genomic ancestry (e.g. ASIP, KITLG), or real effects too small to detect.

“the failure to replicate previously reported effects could be due to a paucity of allelic variation in the Cape Verde ancestral populations (e.g. MC1R), incomplete correction for genomic ancestry with small panels of ancestry informative markers in the earlier studies (e.g. ASIP, KITLG), or real effects of these loci that are too small to detect in our study.”

May combine two claims.

OCA2 was originally considered a strong candidate gene for Afro-European pigmentary variation based on allele distribution patterns of closely linked SNPs and because of a strong association with blue vs. brown eye color in populations of European ancestry.

“OCA2 was originally considered a strong candidate gene for Afro-European pigmentary variation based on allele distribution patterns of closely linked SNPs [11], [14], [34], [35], and because of a strong association with blue vs. brown eye color in populations of European ancestry [4], [41].”

• cites [11] MD Shriver, 2003, Skin pigmentation, biogeographical ancestry and admixture mapping, Hum Genet sourceNot checked yet

  Not checked in the free run.

• cites [14] HL Norton, 2007, Genetic evidence for the convergent evolution of light skin in Europeans and East Asians, Mol Biol Evol sourceNot checked yet

  Not checked in the free run.

• cites [34], [35]Not checked yet

  Not checked in the free run.

May combine two claims.

Previous genome-wide studies of human skin and eye color have focused on populations of European ancestry and have been based primarily on categorical and subjective assessment of pigmentary phenotype.

“Previous genome-wide studies of human skin and eye color have focused on populations of European ancestry and have been based primarily on categorical and subjective assessment of pigmentary phenotype [4]–[7].”

• cites [4]–[7]Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

Hardy-Weinberg equilibrium was not used as a quality filter due to the admixed nature of the subjects.

“Hardy-Weinberg equilibrium was not used as a quality filter due to the admixed nature of the subjects.”

Classified as probably not a substantial claim.

Lamason et al. estimated an effect size for a single SLC24A5 allele in an African-American sample that corresponds to 4.8 melanin units.

“compared to 4.8 melanin units estimated by Lamason et al. [13] in an African-American sample.”

• cites [13] RL Lamason, 2005, SLC24A5, a putative cation exchanger, affects pigmentation in zebrafish and humans, Science sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

SNPs with genotyping call rate <95%, replication errors across 20 replicated samples, apparent heterozygosity equal to one, and only a single allele called in the Cape Verde dataset were removed from the analyses.

“SNPs with genotyping call rate <95%, SNPs showing replication errors across 20 replicated samples, SNPs in which the apparent heterozygosity was equal to one, and SNPs for which only a single allele was called in the Cape Verde dataset were removed from the analyses.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

If a chromatin loop does form between APBA2 and OCA2, the BP3 region must be contained within that loop.

“If, however, a chromatin loop does form between APBA2 and OCA2, the BP3 region must be contained within that loop.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Identity-by-states revealed 12 pairs of cryptic relationships (IBS>0.8); one individual from each pair was excluded.

“Identity-by-states (IBS), estimated using all autosomal SNPs in PLINK [47], revealed 12 pairs of cryptic relationships (IBS>0.8). One individual from each pair was excluded.”

• cites [47] S Purcell, 2007, PLINK: a tool set for whole-genome association and population-based linkage analyses, Am J Hum Genet sourceNot checked yet

  Not checked in the free run.

Classified as probably not a substantial claim. May combine two claims.

SLC24A5 and KITLG have been reported to have effect sizes that correspond to 25%-38% and 20%-25% of the European-West African difference in skin color, respectively.

“SLC24A5 and KITLG have been reported to have effect sizes that correspond to 25%–38% and 20%–25% of the European-West African difference in skin color, respectively [13], [15].”

• cites [13] RL Lamason, 2005, SLC24A5, a putative cation exchanger, affects pigmentation in zebrafish and humans, Science sourceNot checked yet

  Not checked in the free run.

• cites [15] CT Miller, 2007, cis-Regulatory changes in Kit ligand expression and parallel evolution of pigmentation in sticklebacks and humans, Cell sourceNot checked yet

  Not checked in the free run.

May combine two claims.

Admixture stratification is important because the very same loci that underlie trait differences in ancestral populations are also among the strongest admixture informative markers; therefore, incomplete correction for genomic ancestry will lead to overestimation of the effects for any locus that itself is an admixture informative marker.

“Admixture stratification is important in this context because the very same loci that underlie trait differences in ancestral populations are also among the strongest admixture informative markers; therefore, incomplete correction for genomic ancestry will lead to overestimation of the effects for any locus that itself is an admixture informative marker.”

May combine two claims.

The critical SNP rs12913832 is part of an enhancer that forms a chromatin loop with the OCA2 promoter, and the derivative allele causes reduced recruitment of the chromatin remodeler HLTF, leading to reduced binding of additional transcription factors and impairment of loop formation and OCA2 expression in cultured human melanocytes.

“Very recently, Visser et al. [33] showed that the critical SNP in this region, rs12913832, is part of an enhancer that forms a chromatin loop with the OCA2 promoter, and that the derivative allele causes reduced recruitment of the chromatin remodeler HLTF, leading to reduced binding of additional transcription factors and impairment of loop formation and OCA2 expression in cultured human melanocyt”

• cites [33] M Visser, 2012, HERC2 rs12913832 modulates human pigmentation by attenuating chromatin-loop formation between a long-range enhancer and the OCA2 promoter, Genome ResNot checked yet

  Not checked in the free run.

May combine two claims.

Hypopigmentation associated with Prader-Willi and Angelman syndromes is thought to be caused by hemizygosity for OCA2 and is therefore independent of any potential cis-acting regulatory region in APBA2.

“Hypopigmentation associated with these conditions is thought to be caused by hemizygosity for OCA2 [46] and is therefore independent of any potential cis-acting regulatory region in APBA2.”

• cites [46] RA Spritz, 1997, Hypopigmentation in the Prader-Willi syndrome correlates with P gene deletion but not with haplotype of the hemizygous P allele, Am J Med Genet sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

The ∼1 Mb interval between HERC2 and APBA2 contains a cluster of segmental duplications that serves as a frequent deletion breakpoint (BP3) involved in Prader-Willi and Angelman syndromes.

“the ∼1 Mb interval between HERC2 and APBA2 contains a cluster of segmental duplications that serves as a frequent deletion breakpoint (BP3) involved in Prader-Willi and Angelman syndromes [45].”

• cites [45] AJ Makoff, 2007, Detailed analysis of 15q11–q14 sequence corrects errors and gaps in the public access sequence to fully reveal large segmental duplications at breakpoints for Prader-Willi, Angelman, and inv dup(15) syndromes, Genome Biol sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

Two samples with call rate below 95% were removed.

“Two samples with call rate below 95% were removed.”

May combine two claims.

We favor the alternative that many genes of small effect cumulatively explain the correlation between skin color and individual ancestry, and together with several genes of moderate effect (GRM5-TYR, APBA2 (OCA2), SLC45A2, and SLC24A5), account for high overall heritability.

“We favor the latter alternative, which supports a view of skin color genetic architecture that is multimodal, in which many genes of small effect cumulatively explain the correlation between skin color and individual ancestry, and together with several genes of moderate effect (GRM5-TYR, APBA2 (OCA2), SLC45A2, and SLC24A5), account for high overall heritability.”

Classified as probably not a substantial claim.

The extracted irises were manually inspected and found sufficiently accurate.

“The extracted irises were manually inspected and found sufficiently accurate.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Age was found not correlated with the phenotypes (P>0.5 for both skin and eye colors), and hence was not included as covariate.

“age was found not correlated with the phenotypes (P>0.5 for both skin and eye colors), and hence was not included as covariate.”

May combine two claims.

Effect sizes shown in Figure 5 for KITLG (0.19 MM-0.26 MM) are estimated from Miller et al. in which the beta value for rs642742 is reported as 3.8 melanin units based on a linear regression model, and 2.8 melanin units based on ADMIXMAP.

“Effect sizes shown in Figure 5 for KITLG (0.19 MM–0.26 MM) are estimated from Miller et al. [15] in which the beta value for rs642742 is reported as 3.8 melanin units based on a linear regression model, and 2.8 melanin units based on ADMIXMAP.”

• cites [15] CT Miller, 2007, cis-Regulatory changes in Kit ligand expression and parallel evolution of pigmentation in sticklebacks and humans, Cell sourceNot checked yet

  Not checked in the free run.

May combine two claims.

Population-attributable variance is non-additive; variance accounted by two loci is not the sum of the PAV of the two loci individually.

“We note that the PAV is non-additive; in other words, variance accounted by two loci is not the sum of the PAV of the two loci individually.”

May combine two claims.

We also carried out an association analysis with EMMAX, which adjusts for population stratification by including a relationship matrix as a random effect; the results were similar to those obtained using conventional association analysis.

“We also carried out an association analysis with the program EMMAX [22], which adjusts for population stratification by including a relationship matrix as a random effect; the results (Figure S1) were similar to those obtained using conventional association analysis (Figure 3).”

• cites [22] HM Kang, 2010, Variance component model to account for sample structure in genome-wide association studies, Nat Genet sourceNot checked yet

  Not checked in the free run.

Classified as probably not a substantial claim.

To exclude the possibility that the association between skin color and genome-wide ancestry is due to geographic confounding between islands, we examined this correlation within each island.

“To exclude the possibility that the association between skin color and genome-wide ancestry is due to geographic confounding between islands, we examined this correlation within each island (Table S1).”

• AssumptionThe authors state this as an assumption: The Cape Verdean population has two ancestral populations (European and West African).

Probably not a substantial claim: may not pass S3 (load-bearing).

The effect size shown in Figure 5 for ASIP (0.24 MM) is estimated from 3.5 melanin units based on data in Bonilla et al.

“Similarly, the effect size shown in Figure 5 for ASIP (0.24 MM) is estimated from 3.5 melanin units based on data in Bonilla et al. [12].”

• cites [12] C Bonilla, 2005, The 8818G allele of the agouti signaling protein (ASIP) gene is ancestral and is associated with darker skin color in African Americans, Hum Genet sourceNot checked yet

  Not checked in the free run.

May combine two claims.

While CEU and YRI are imperfect ancestral populations, they do not cause a large bias in either genome-wide or local ancestry estimates.

“Thus, while the CEU and YRI are imperfect ancestral populations, they do not cause a large bias in either genome-wide or local ancestry estimates.”

• AssumptionThe authors state this as an assumption: The Cape Verdean population has two ancestral populations (European and West African).

May combine two claims.

The genetic relationship matrix based on allele sharing is inflated by population structure; in addition, pilot studies observed a large sampling variability of GCTA-estimated heritability due to the moderate size of the Cape Verde sample.

“However, the genetic relationship matrix based on allele sharing is inflated by population structure [55]; in addition, in pilot studies we observed a large sampling variability of GCTA-estimated heritability due to the moderate size of the Cape Verde sample.”

• cites [55] SR Browning, 2011, Population structure can inflate SNP-based heritability estimates, Am J Hum Genet sourceNot checked yet

  Not checked in the free run.

Classified as probably not a substantial claim. May combine two claims.

Assessment and control for population stratification is described in Results; the P values reported in Table 1 are derived from linear regressions using PLINK in which the first 3 principle components and sex are included as covariates.

“Assessment and control for population stratification is described in Results; the P values reported in Table 1 and are derived from linear regressions using PLINK in which the first 3 principle components and sex are included as covariates.”

• AssumptionThe authors state this as an assumption: The Cape Verdean population has two ancestral populations (European and West African).

May combine two claims.

The simulation approach preserves a realistic level of correlation structure between phenotype, genome-wide ancestry proportions and genotypes, and takes into account the two strongest predictors of phenotype: genome-wide ancestry and genotype at SLC24A5.

“This approach preserves a realistic level of correlation structure between phenotype, genome-wide ancestry proportions and genotypes, and also takes into account the two strongest predictors of phenotype: genome-wide ancestry and genotype at SLC24A5.”

• AssumptionThe authors state this as an assumption: Residual non-genetic variance is independent of any genotype or ancestry.
• AssumptionThe authors state this as an assumption: The simulated phenotype has a genetic architecture with three major loci, each with the same effect as SLC24A5 and perfectly ancestrally informative.

Classified as probably not a substantial claim. May combine two claims.

Imputed genotypes were inferred with MaCH 1.0.16 using combined CEU and YRI haplotypes as reference; we report SNPs whose predicted R∧2 between true and imputed genotypes >0.80.

“Imputed genotypes were inferred with the MaCH 1.0.16 package [52] using combined CEU and YRI haplotypes as reference; we report SNPs whose predicted R∧2 between true and imputed genotypes >0.80.”

• cites [52] Y Li, 2010, MaCH: using sequence and genotype data to estimate haplotypes and unobserved genotypes, Genet Epidemiol sourceNot checked yet

  Not checked in the free run.

Classified as probably not a substantial claim. May combine two claims.

We restricted the association scans to the 879,359 autosomal SNPs with MAF>0.01; SNPs achieving a P<5.7×10−8 were considered genome-wide significant.

“We restricted the association scans to the 879,359 autosomal SNPs with MAF>0.01; SNPs achieving a P<5.7×10−8 were considered genome-wide significant.”

May combine two claims.

In simulation studies, the ARHMM achieves comparable accuracy as HapMix, but is more flexible and does not require information about the recombination rate.

“In simulation studies, the ARHMM achieves comparable accuracy as HapMix [49], but is more flexible and does not require information about the recombination rate.”

• cites [49] AL Price, 2009, Sensitive detection of chromosomal segments of distinct ancestry in admixed populations, PLoS Genet sourceNot checked yet

  Not checked in the free run.

May combine two claims.

If G1 and G2 represent derived (pigmentation lightening) alleles at two loci, then cov(G1, G2)>0 due to admixture linkage disequilibrium, γ1γ2>0, and V12<(V1+V2).

“In our situation, if we code G1 and G2 to represent the derived (pigmentation lightening) alleles at the two loci, then cov(G1, G2)>0 due to admixture linkage disequilibrium, γ1γ2>0, and we have V12<(V1+V2), as is indicated by the overlapping areas in Figure 7.”