Fig 1.
A) Manhattan plot of associations with blue vs. brown eyes across the genomes of 3,180 dogs. Horizontal lines represent the thresholds for suggestive (grey; P < 1x10-5) and significant (black; P < 5x10-8) associations. B) Read depth (scaled by the average depth across the interval for each dog) in 10-kb sliding windows across the CFA18 GWAS peak region, for the six Siberian Huskies with publicly available whole genome sequence data (blue) and 11 dogs from other breeds (grey). Five of the six huskies and five of the 11 other breeds carry the GWAS allele associated with blue eyes (dot at 44,924,848). Black vertical lines indicate paired-end reads that aligned 98.6-kb from their mate and in an opposite orientation. Photo credit: Aleksey Gnilenkov (Flickr).
Fig 2.
PCR genotyping of a tandem duplication upstream of ALX4 associated with blue eye color.
A)* Schematic view of brown- and blue-eyed alleles (not to scale). The duplication sits head to tail to the ancestral sequence. Three sets of primers were used to amplify three regions (primers denoted with single headed arrows). Sanger sequencing of the duplication midpoint show nearly perfect synteny to canFam3.1 chr18:44791409–44791553 and 44890066–44890185. A single basepair difference, highlighted in red, show a T in the duplication sequence that corresponds to a G at chr18:44791413 in the ancestral sequence. B) PCR genotyping of one brown-eyed and one blue-eyed dog. Primer pairs denoted above each PCR lane. The 5' and 3' flanking regions amplify in both the brown- and blue-eyed alleles; the duplication midpoint amplifies only in the blue-eyed allele.
Fig 3.
Scaled density plot of Δ log R distributions for discovery panel dogs with zero, one, or two copies of the associated haplotype, demonstrating that the presence of the haplotype tracks the presence of the duplication in almost all cases.
Dogs carrying the haplotype exhibited elevated log R at SNPs within the duplicated region compared to flanking regions (high Δ log R) relative to non-carriers, and dogs heterozygous vs. homozygous for the haplotype exhibited distinct distributions, consistent with being heterozygous vs. homozygous for the duplication itself. Although the duplication appeared to act dominantly in Siberian Huskies, brown-eyed heterozygotes in other breeds or mixed breed dogs also had log R data consistent with carrying the duplication. Exceptions included three high-log R dogs with alternative, recombinant versions of the associated haplotype (asterisks, top panel; S7 Fig) and one low-log R dog with a partial duplication (asterisk, middle panel; Supplementary Information). Individual log R values contributing to each density curve are represented with vertical ticks, and counts of blue-eyed vs. brown-eyed dogs are indicated for each haplotype category.