Hi Rabia,

The most obvious question that comes to my mind is whether or not the probability of occurrence of hip displasia is driven by the genetics of the dog breed.

The idea would be to have a phylogenetic tree of all existing breeds, including wolves, on top which you would map, as a trait, the occurrence of hip displasia (obtained from vets around the world for instance).

Two possibilities. If hip displasia is strongly driven by the phylogenetic tree, it means that certain breeds that are genetically closely related will tend to have the same probability of occurrence of hip displasia (whether low or high). Inversely, if it is not driven by the phylogenetic tree, you will have the different probabilities randomly distributed across the tree. And of course, everything between these two extremes.

Assume that you have molecular markers for all the dog's breed (which you can probably find already published), you have many different ways to draw the phylogenetic tree. I would suggest to use BEAST v2 (Bouckaert et al 2014: doi:10.1371/journal.pcbi.1003537). Then, to test for the significant effect of hip displasia (as a trait) on the phylogenetic trees there are also several possibilities.The program BaTS 2.0 (Parker et al., 2008: DOI:10.1016/j.meegid.2007.08.001) for trait analysis can do the job.

Good luck for the assignment

Thierry B Hoareau Thank you for that, it was really helpful

Following Thierry B Hoareau 's suggestion, a pretty commonly used statistic for testing for phylogenetic signal in trait data is the K statistic from Blomberg et al. 2003 (attached), which basically asks if the given tree better explains your trait data (=incidence rates of hip dysplasia) than if we randomly permuted the trait data on the tree. If there if phylogenetic signal (=related individuals have more similar traits), then the fit of the tree to your data should be much better than to the random permutations. It wouldn't be too hard to do this test- you will need to create a table 'coding' the tips with your trait data:

e.g.

Breed hip_prob

Dalmation 0.12

Labrador 0.18

...

...

And you can use an R package called 'picante' to perform the test with only a few lines (from http://picante.r-forge.r-project.org/picante-intro.pdf):

#first install packages:

install.packages(c("ape", "picante"))

#then do the analysis:

library(ape) #loads packages

library(picante)

tree