Originally Posted by csagan001  Here is what I learned so far:
The lab has tested “at least” 25 dogs for each of the markers they claim cause disease in specific breeds. (Note that in doing similar testing on humans, pharmaceutical labs do the tests thousands of times.) 25 dogs is less than that required for minimal statistical analysis by at least 17%. As a result, their statistics cannot be relied upon for decision making. In other words, their error rates are far too high for any useful conclusions regarding any specific dog to be made.
The DNA studies completed so far, and there have been surprisingly few of them are looking for “markers” only. A marker is not what most people think it is. The canine genome consists of 2.4 billion “letters” (each letter is the “name” of a particular protein). A marker is only an indicator that a specific sequence of letters, within one of the dog’s 39 chromosomes is correlated with a particular trait.
For example consider this: I am going to buy a used car. I know that the used car has 75,000 miles on it, and I can see that it is highly polished and has good tires. I cannot see the inner workings of the engine or the transmission. However, I can take it for a compression test. The compression test will let me know the health of the engines cylinders and its valves. So, I have the compression test done and discover that one of the cylinders has a different reading from the others. From this information I determine that the engine MIGHT need some costly repairs in the near future. However, what I did realize was that the compression test was given to a car that uses various numbers of cylinders, depending on the work load asked of the engine (yes, there are cars like this—Honda calls it i-VTEC—it saves lots of gas. The compression test is nothing but a marker. Depending on the car it is used on, the testing methods, and other factors, it may give information useful to the used car buyer. However, in this example, it gave the used car buyer information that led to an incorrect conclusion.
Ok, using this same idea for dogs—If I have a great Dane, and I research its DNA for longevity characteristics, I may come to the conclusion that a specific “marker” is present for great Danes that live longer than others. This marker may not be the cause of long lives, it may merely be present for a majority of dogs that live a long time. Since longevity is not caused by a single gene—that is, it is caused by the relationships of many genes (term: “polygenetic”)—the marker does not tell the whole story. It is merely the indicator that “long life genes” may be present.
If you then look for this gene in a completely different breed, say a Maltese, I might find it. Does that mean that the Maltese with the marker present will live longer? Answer: No. Remember: longevity is based on many genes working together in a specific fashion, and that the marker we found in our original research was based on great Danes. To utilize the marker in Maltese means that we need to test a bunch of Maltese dogs and cross compare the results. Since doing this takes time, and costs money, the results are slow in coming.
All that is fine and good, but what does it tell us about using a DNA test as an indicator of probability of disease in the future, for a beagle, if the test was developed on a schnauzer? Answer: Nothing. A test developed on one breed is largely useless for other breeds unless it can be statistically validated.
The company has not statistically validated their tests for any breeds except the ones that they claim. So, the tests that are being conducted on the Biewers are completely bogus. They cannot be used to ascertain the probability of any specific dog becoming ill. As a matter of interest, the tests should not be used on the breeds for which they have been validated. This is because they only tested 25 dogs of each breed. They need to test at least 900 dogs (that would give them 30 groups of 30 dogs each….enough for a valid statistical and very meaningful study.) |