Before embarking on the Georgie Project we had to decide if it was feasible.  Did we have access to the necessary materials?  Would it be too expensive?  The first 18 months of the project were concerned with these issues.

We obtained pedigrees going back to the dogs which founded the current Portuguese Water Dog population.  We constructed a new computer program--Georgie--which allows us to investigate the genetics of large populations.  With more than 8000 animals in the computer we were ready to find out some answers.

One of our primary concerns was the accuracy of the pedigrees.  We were going to use the pedigrees to establish the feasability of the study.  To test their accuracy, we genotyped a number of dogs and measured their relatedness from these genotypes.  We also calculated their relatedness from the pedigrees.  The agreement was excellent, indicating that the pedigrees were accurate and that we had not made many mistakes in our genotyping.

We then used the actual pedigrees in computer simulations to determine how many dogs we would need and how many genetic markers in order to find the genes that control complex genotypes.
 

Here are the answers:
 

Limiting the number of genetic markers and the number of dogs is important.  It costs approximately $2/dog/marker to measure genetic markers commercially.  In a university research lab the cost can be reduced to about $1.25/dog/marker.

Clearly it was impossible to genotype more than 8000 dogs.  Our analysis told us that we needed a minimum of 400 dogs and 600 would be much better.  With 400 markers per dog, we estimate a cost for genotyping between $200,000 and $300,000.

We have the resources to do this!!!

In cooperation with owners and breeders, we have started to collect blood for DNA from a subpopulation of dogs and now have more than 400.  The sub-population has been chosen on the basis of four criteria:
 

Thanks to the work of others in the field, most particularly Dr. Elaine Ostrander,  we have available a large number of molecular markers on the canine genetic map which we can use for Portuguese Water Dogs.  At present we are using more than 200 of these.

Before we could embark on the project, we had to determine whether the anatomical phenotypes of  Portuguese Water Dogs were suitable for a study using as few as 400 dogs.  If genes determined relatively little anatomical variation compared to the changes produced by diet or exercise and by variations in measurement, we would not be able to identify them.

This is illustrated by the figure below in which the dots in the background are random environmental factors or measurement errors and the genes for CHD are either little or big dots.  When the genetic effect is big (big dots), we have no trouble identifying it.

As an initial survey of the Portuguese Water Dog population, we let our collaborators, the owners or breeders, make the measurements using a measuring tape and a stick.  Surveys were sent to 500 owners from which we received reponses with measurements on 365 dogs.  From these responses we determined that the genetic signal was strong enough to read.