How we know Wiston Cap carried Collie Eye Anomaly

In a post from 2010, Through Anomalous Eyes, I exposed how statistical analysis proves the long rumored–and often denied–belief that Wiston Cap, the most popular sire in Border Collie history, carried the gene for the recessive disease known as Collie Eye Anomaly.  Using the same statistical analysis, it’s also assured that Cap (the second most popular sire), was also at least a carrier if not affected with CEA.

International Supreme Champion Wiston Cap (ISDS 31154) was certainly a carrier of Collie Eye Anomaly.

How can we be so sure International Supreme Champion Wiston Cap was a genetic carrier for Collie Eye Anomaly if he was long dead before there was a test for CEA?

We can’t test Wiston Cap directly, he’s dead and gone. But Wiston Cap and Cap have left us a DOCUMENTED genetic legacy via the stud book. We know tens of thousands of their descendants and we know HOW those dogs are descended from Wiston Cap and Cap. Even with pedigree fraud and error, the volume of accurate pedigrees and the growing number of dogs being tested combine to inform what genes Wiston Cap had and passed on.

Every Border Collie tested today, and they’re all related to Wiston Cap and Cap, tells us a lot about that dog, a good deal about that dog’s parents, some about their grandparents, etc. and a little something about Wiston Cap and Cap.  The more dogs tested, the more accurate the picture we have of Wiston Cap and Cap.

We KNOW that Wiston Cap was a carrier for CEA because the pattern of affected and carrier dogs which HAVE been tested are more than 99% consistent with Wiston Cap being a carrier.  There is a ~0% chance that Wiston Cap was affected and a ~0% chance that he was clear.  If he were affected we would expect a very different pattern of disease in his offspring and their offspring and their offspring; specifically we would expect to see a lot more dogs being tested as carriers and affected than we do.  Likewise, if he was clear, we would not expect to see the pattern of affected and carrier dogs that we do see.

How can we know what an ancestor had without testing that ancestor?

We can test their descendants and use pedigree analysis to reconstruct with great accuracy what the ancestor’s genes looked like.  Like a Sudoku puzzle, we use incomplete information and deduction to reconstruct information that isn’t obvious when we start because we know that there are rules that were followed as the information was passed down to us.  If we start at the mouth of the Mississippi River at its delta near New Orleans, we might think we could never figure out where all this water came from, but we can. We can start going back up-stream and finding out where the tributaries enter.  If we test the water at the mouth and find there’s some toxin in it, we can even deduce where the toxin is coming from without testing all the thousands of tributaries, we can start testing concentrations as we go back upstream and find where the signal is stronger, where it’s weaker and where it’s absent and that will allow us to paint a picture of what upstream must look like.

Wiston Cap’s genes flow down to us through hundreds of streams that are defined and recorded by the stud books.  The more of these streams that we test, the more we can tell about Wiston Cap until the odds stack up and we are more certain of what Wiston Cap had than we even would be if he were tested himself (the certainty of our calculations can surpass the error rate of the test!).

How do we establish certainty about ancestors like Wiston Cap by only testing current dogs?

For some genes, we know near exactly what an ancestor had without testing them, as long as we do test one of their descendants. Your mitochondrial DNA is the same as your mother, and her mother, and her mother before her (save for the rare mutation). This mtDNA is what allows current scientists and genealogists to tell you where some of your ancestors were 20,000 years ago. For men, their Y chromosome is the same as the one their father has, and his father, and his father. Like mtDNA, we can say with great certainty that yDNA your male line ancestors had. We don’t need to test them, we know how this works.

In dog breeding, we call the y-chromosome line the “Sire Line” and the mitochondrial ancestry is represented by the “Dam Line.”  You can learn more about the popular sire lines in Border Collies in my post: Popular BC Sire Lines.

If we tested a number of direct male-line descendants of Wiston Cap, we would be able to tell with high certainty what his y-chromosome looked like (and we could even weed out pedigree fraud if we found a minority of claimed male-line descendants who did not share the same y-chromosome).  We don’t need to test Wiston Cap directly because we have thousands of offspring to test and the more of them we test the more we know we’re certain about Wiston Cap.

CEA is not sex-linked (mutation occurring on the y chromosome or the x chromosome) nor a mitochondrial disease, so this type of certainty does not apply to CEA but it does apply to other diseases that are passed on with a sex-linked or mitochondrial pattern, so it’s worth mentioning. You can brush up on the difference between mtDNA, yDNA, and Autosomal DNA in my post: The 3 Types of Genetic Inheritance.

How do we logic our way to certainty about ancestors like Wiston Cap for genes that aren’t always passed down to every offspring? 

Not all genes are deterministic like y chromosomes and mitochondria. For most of your DNA, you get one allele from your mother and another from your father. Both of your parents had two alleles from which to give you their one, so there’s some uncertainty.  Still, this does not prevent us from reconstructing what alleles they had to pass on, especially when we have multiple children and multiple generations of information.

A little bit of knowledge can be a dangerous thing, but a lot of data can really clarify the situation.  For example, let’s begin with total ignorance.  We don’t know what our dog has or does not have. Our dog can be in one of 3 states: CEA CLEAR, CEA CARRIER, CEA AFFECTED.  Without knowing anything about the breed-wide distribution of CEA, we’re really hard pressed to say which one of those states is the most likely for our dog.  The breed could be devoid of CEA and if we only knew this we’d be able to say without testing that he was clear.  The breed could be saturated, and in that case we’d be certain that he was affected.

When we don’t know which of 3 states our dog is in {Clear, Carrier, Affected} we don’t know which of 9 states the parents are in: { (father clear, mother clear), (father clear, mother carrier), (father clear, mother affected), (father carrier, mother clear), … , (father affected, mother affected) }.

So we test. We get a CLEAR result.  What can we say about the parent dogs?

Well, we can say with CERTAINTY that neither parent is affected. So (father affected, mother affected) now has a 0% probability.  We can also say that any of the 9 possible states that include an affected parent are 0% probability too.  Why? Because if either or both parents are affected the offspring could not be clear.  So of our 9 possibilities, we’ve ruled out 5 of them as impossible. We know that each parent is now either Clear or Carrier.  We can assign probabilities to these, but I’ll spare you the math.

So this one test result has told us some things with certainty about the parents without testing them and other things we can apply probabilities.

As we test more offspring of these dogs (together or with other mates) we can adjust the probabilities that they are Clear or Carrier.  The more puppies that either one produces that are clear, the higher percent we can assign to them being clear.  If they ever produce an affected, we KNOW with certainty that they are a Carrier, and so is the other parent they produced an affected puppy with.

Likewise, the more dogs we test, the more we can with certainty tell what their parents would test, and the more parents we nail down with certainty, the more we can pin down the grandparents.  First we narrow the probabilities and then, sometimes, we can pin them down with certainty too.  Even without logical certainty, the more information we have the more our combined probabilities approach certainty.

So what happens when we get a CARRIER result? What do we know about the parents? 

When we know nothing else about a dog’s parents, when it comes up a CARRIER for a recessive disease we can say there are several possibilities. We know that there are 9 total parent states, and a carrier result rules out two of them (father clear, mother clear) and (father affected, mother affected).  Neither of those two states could produce a puppy that is a carrier.

For example, if all of the children of a sire are carriers for CEA and none of them are affected, we are pretty sure that the sire is affected and all the dams are clear.  Every additional child of that sire that comes up a carrier or affected makes us more certain of this.  If any child came up clear we would have to reevaluate the odds we placed on affected and carrier.  A clear child should not come up from an affected sire, so either all the other children are a statistical fluke or the sire is not actually the sire we think.

If we have a sire who has bred with multiple females, as we do in dogs and certainly as we have with Wiston Cap and Cap, we can begin to distinguish which parent is responsible for the pattern we see in their descendants.  For example, should either parent produce an affected, we KNOW that each parent is AT LEAST a carrier.  Neither one can be clear and produce an affected.  If a parent never produces a clear, we’re certain that they are more likely to be affected than a carrier.

If one parent is affected, every single one of their offspring will be at least a carrier.  So if we find that for an untested parent every tested child is a carrier and some are affected, the odds increase that the parent is affected vs. carrier.

The same logic can be used when a dog comes up affected.  We know that he got a copy from both parents so both are at best carriers, perhaps affected themselves. Looking at other offspring will begin to paint the picture of what the parents actually are.

Does the testing only tell us about untested parents or does the information reach back further in the pedigree?

There’s no limit on how many generations back a modern test will inform, but the ability of any single test this generation does fall off by a power of 2 each generation.  For example, if we only test one dog, and thus have no means to differentiate between the parents, we assign each parent the same probabilities of being clear, carrier, affected based on what the dog we tested came out.  The more puppies we test, the more we can affirm or change those percents.  But as soon as we assign any probability to a parent, we can assign a probability to their parents.  And so on.

A single test only shifts our default probabilities so much, but multiple tests across a variety of descent paths really do move the probability matrix back across multiple generations, often approaching near certainty.  Because Wiston Cap was bred a lot we have a lot of data downstream to create an accurate picture of what he was.

Every new test makes all the previous information in our probability network more accurate.  For example, if we start with only one puppy from a sire x dam being tested we can’t really differentiate the probabilities between the sire and dam. But then the dam has a new litter and several of those puppies are tested and we get a much better picture of what she is like.  This information will flow back upstream to her previous litter and will change what probabilities we assign to that sire.  Likewise, when he is bred again, and some of those descendants are tested, the information will improve our picture of him and of all his previous mates and then all their children, etc.

How can this type of analysis provide even MORE certainty than a direct DNA test?

Every test, even DNA tests, have an error rate: the results can read affected when the truth is clear.  This is why doctors will retest for important results even with tests that are 99%+ accurate.  Pedigree analysis, the sort of deduction I have been explaining, can even provide higher certainty than a direct DNA test.  This happens because one single DNA test might have a 98% accuracy rate… it only returns a false result 2% of the time.  But when a dog breeds and passes along its genes, it’s performing an event which reveals something about its makeup.  The more puppies it produces and the more puppies they produce in a documented manner, the more likely their genetic profiles will mirror the expected outcomes.

As we test all those puppies, we are at the same time performing a test on the parent, just not a direct test, but the results are still informative.  If we only perform 1 test on the parent, our results will be wrong 2% of the time.  But if we perform hundreds of tests on the children, even though the results have only half the power individually to inform the parent, the sheer number of tests makes the collective error rate less than 2%.

Can we be as certain of a derived result as we can be of a direct one?

Yes.  Because everything is actually a derived result if you think about it.  When you get your test back from the lab, do you KNOW that it’s accurate?  Or could that be the two in a hundred that are accidentally wrong?  What if they tested multiple times?  Well we can be MORE certain but not perfectly certain.  A dog who is just a carrier could test once, twice, three times as affected.  It could also, by sheer chance, always pass along its affected allele and never its clear allele and thus it could even breed like an affected dog instead of a carrier.  All possible.  Just not PROBABLE.  So even if we get that dog tested three times and he produces a dozen puppies, there is a non-zero chance we’re wrong.

But we don’t need to live life like this… it’s a very very very small chance.  So we can move along and will likely never have that come back around to bite us.  We don’t need to stress about remote possibilities.  And the fact that they always exist doesn’t corrupt the rest of our decision-making either.  I mean the Earth could end tomorrow, and if we KNEW that, you could probably skip work today and go have a last fling before the end.  But it’s not likely, so if you keep skipping work and going on flings on the remote chance that the earth will be gone tomorrow, you’re probably going to screw up your life for all the much more likely outcomes.

So yes, there’s some remote certainty that Wiston Cap was not a CEA carrier. But it’s no larger than the remote chance that if he were directly tested that he’d come up clear.  And given the entirety of the thousands and thousands of litters since that have come up with CEA based on the exact pattern we’d expect if he was a Carrier, there is NO REASON to believe anything other than that Wiston Cap was a CEA carrier.  The totality of the facts say only that one thing.  All of the testing that has been done crossed with all of the pedigrees that have been kept make Wiston Cap being a carrier as much or more certain than if we could have him tested directly right now.

So if anyone denies this, they are doing so because they don’t want it to be true, not because it’s at all a reasonable position to hold and not because it’s consistent with any facts we have.

* * *
Comments and disagreements are welcome, but be sure to read the Comment Policy. If this post made you think and you'd like to read more like it, consider a donation to my 4 Border Collies' Treat and Toy Fund. They'll be glad you did. You can subscribe to the feed or enter your e-mail in the field on the left to receive notice of new content. You can also like BorderWars on Facebook for more frequent musings and curiosities.
* * *