This web page was produced as an assignment for an undergraduate course at Davidson College.

"Coat Variation in the Domestic Dog is Governed by Variants in Three Genes"

Cadieu et al.

The paper: its ideas and claims

The researchers who produced this journal article were interested in loci involved in domestic dog coat phenotypes including growth pattern, length, and curl variation. They designed their data collection based on the idea that the use of an individual breed will allow for targeted variation to locate the locus of specific phenotypes, and the use of multiple breeds to locate the specific gene responsible for these phenotypes. They researchers go on to say that this strategy could give insight into other complex phenotypes.
            The focus of this paper was locating three loci, one for each of the phenotype variation in question. The three phenotypes they investigated were hair length, presence of absence of mustaches and eyebrows (furnishings), and presence or absence of curls. In order to investigate these loci, they produced three genome wide single nucleotide polymorphism (SNP) data sets. The first data set used dachshunds with three varying phenotypes of furnishings. The second data set used Portuguese water dogs to investigate varying curl phenotypes, and the last data set, what they call CanMap, included 903 dogs from 80 breeds with many different hair phenotypes.
            The first two data sets were the loci finding sets. The researchers did what is called a genome-wide association study (GWAS) which produces data indicating relationships between SNPs and phenotypes. The third data set was used to rule out what they call false positives, a locus that seems to be linked in one breed but not in other breeds. After the GWAS, the researchers mapped the strongly linked loci and sequenced specific haplotypes in order to identify the mutations linked to each phenotype.
            The researchers found very compelling support that there are three genes encoding R-spondin-2, firbroblast growth factor-5, and keratin-71 (RSPO2, FGF5, and KRT71). They provide evidence that RSPO-2 is directly related with hair growth and is a great candidate for the furnishings gene. They used information from previous experiments to support the claims that FGF5 and KRT71 are associated with long hair and curly hair respectively. The logic and data they provide for these claims are very well explained and supported by their supplementary data and other studies.

            Overall, this journal article is well organized and logically makes sense. Not once did I see the researchers stretch their claims because of over interpreting data. Other than minor details in this journal, I saw no significant problem with how they presented their data. On a more personal note, I also liked their introduction and reason for conducting these experiments.


The figures: what they show and what they mean

Figure 1.

Figure 1 begins with three pictures of common dachshund coat phenotypes. This part of the figure is quite self-explanatory and is meant to give some context to the three phenotypes the researchers list, where smooth coated and long haired dogs do not have furnishings. Part B of figure 1 is the results for the GWAS for the dachshund breed. In this data set the researchers used wired haired dogs as the test group and smooth coated and long haired dogs as the controls. The arrow points out the locus with the most significant logarithmic of the odds ratio for linkage (lod) score. Figure 1 C is also results from a GWAS. The difference between 1B and 1C is the groups of dogs the data sets were taken from. Figure 1C is a GWAS from the CanMap group. I like the comparison between the two figures because it shows that there are false positives in the dachshund only data set (the other peaks in the reading). Basically, Figure 1C supports and finalizes the assumptions made in figure 1B. Figure 1D are the results of the researchers fine mapping. The red rectangle correlates to the haplotype of the dachshund data group (nice of them to make related data the same color). Predictably, the blue bar represents a homozygous region from 19 breeds fixed for furnishings (Figure 1C). The green rectangle represents a more specific homozygous region between 18 breeds after fine mapping. The red line indicates a deletion that occurs in the above three haplotypes and the boxes below it represent genes in this region of chromosome 13. This part of figure 1 shows that the polymorphism linked to furnishings is found in the 3' untranslated region of the RSPO2 gene.

Figure 2.

Figure 2A shows a similar mapping process used in figure 1D. This mapping was done in search of the gene that affects coat length. Once again, the dachshund(red bar indicating haplotype from 29 daogs) and CanMap (blue bar indicating a homozygous region in 319 dogs from 31 breeds) groups were used. Following the logic from Figure 1D we can see that the most relevant SNP (highest lod score) is found in the FGF5 gene. The last two parts of figure 2 relate to the curly hair phenotype. Before I start on these two figures, I would have liked to see the lod score readout for the length phenotype, but I figure this was left out because of space limitations. Figure 2B shows pictures of the two phenotypes associated with curly vs non-curly coats. As you can see, the researchers used Portuguese water dogs as the intrabreed investigation. Figure 2C shows another mapping sequence. The green bar represents a homozygous haplotype of 65 dogs from five curly haired breeds. The top red line represents the best SNP from the mutibreed data set and the bottom red line represents the best SNP from the Portugeuse water dog data set. The genes in this area are depicted below and one can see that both SNPs are in the KRT71 gene. Figure 2C was a little more difficult to grasp at first, but I think it is cool how specific breeds show variance in the SNP causing the phenotype variation.

Figure 3.

Figure 3 is a simple table that shows the combination of all three genes, their genotypes, and the resulting phenotypes. The minus (-) sign indicates the ancestral variant of each gene, and the plus (+) sign indicates the variation from the ancestral gene. It seems dogs with short hair (A) depict the ancestral phenotype. Pictures of each phenotype can be seen to the right. I am curious as too why they did no include the genotype FGFR5-, RSPO-, KRT71+. The researchers do not really adress this, but the kind of leave the door open by saying "we show that combinations of these genotypes give rise to at least seven different coat types."

Closing Remarks and Referencess:

This article was one of the better genomics articles I have read. The researchers could not include all their data in the space provided so they chose which parts of their study were most valuable to what they were trying to show and described them in a reasonable and concise manner. Figure 2 was the weak figure to me bacause it seemed to be missing some information but upon scrutiny, one can understand what the authors were trying to share with the reader.


Cadieu E, Neff MW, Quignon P, Walsh K, Chase K, Parker HG, Vonholdt BM, Rhue A, Boyko A, Byers A, Wong A, Mosher DS, Elkahloun AG, Spady TC, Andre C, Lark KG, Cargill M, Bustamante CD, Wayne RK, Ostrander EA. Coat variation in the domestic dog is governed by variants in three genes. Science. 2009 Oct 2;326(5949):150-3. : SNPs informational webpage : GWAS informational webpage : lod score informational webpage




My Hompage

Genomics Page
Biology Home Page

Email Questions or Comments to

© Copyright 2011 Department of Biology, Davidson College, Davidson, NC 28035