by Snowleaf on 14 February 2017 - 10:02
Thanks! Love this tool!
by susie on 14 February 2017 - 22:02
That´s a very long time...and even if you are able to trace back 10 generations you will never know which genes ( 50:50 ) were inherited to the next generation. The COI may show the overall percentage, but never the single traits you are looking for.
For instance breeder x bred a litter because of working ability of sire and dam, breeder y bought a puppy out of this litter, but selected for color, not for working ability - out of this litter breeder z bought a puppy, and he selected for size or coat, and so on...
During the same time all these breeders ( 10 generations = thousands of breeders ) not only used one dog, but two ( sire and dam ). That´s a lot of unknown blood...and you will not be able to point out one ancestor for one trait, simply too many variables.
by Snowleaf on 15 February 2017 - 07:02
Thank your for your answer! I'm aware of the data, and of the fact that we're talking probabilities. However, I'm not primarily interested in linebreeding, but in avoiding inbreeding. With Australian Shepherds, we have about 300 foundation dogs, and in the early days there were a lot of linebred dogs and favorite studs. Thus, in order to preserve the genetic vitality and heterozygotism of the present day Aussie, it is recommended to go back ten generations in order to get a clear idea of which dogs to mate. Referring to C.S. Sharp on Ashgi.org here, in case you're interested in reading up on that.
In my case, when I go back five generations, the COI is null. However, in generation 6 and 7 the first common ancestors begin to appear, and if you go back ten generations, you find like 6, 7, 10 times the same dog. I just want to know the ratio for each possible stud in order to select for greatest homozygosity.
As for the mantra of linebreeding: Statistically there is no correlation between success in shows and linebreeding. Historically (in the case of Dalmations, at least, which were used for a study) there were just as many outcrossed winners as there were linebred ones.
by Oli on 15 February 2017 - 19:02
Hi, There is a reason that 7 is the limit. With dogs that have 100% lineage for 10 generations, the system needs to do massive calculations (including power of and squared) for 7,257,600 dogs. Since the last generation (10th) still needs to check 5 generations back, (according to the formula). Of course there aren't that many dogs in the system, but it shows the scale. Of course its the same dog for most parts as you go up, but checks need to be done
Since web crawlers and harvesters show no respect (they are about 93% of all traffic, humans around 7%), I had to limit it to 7 generations and still keep the calculations fairly fast. This is done in real-time, every time the page is viewed (or generations changed).
I could add the ability to go to 10, 15, 20 . but it would have to be extremely limited and only very few persons could use it due to the strain on the database. So, this is possible. I just need to put down heavy guidelines on who can use it and when.
I checked out this dog: Crazy-River-Enjo with 10 generations, it took the system (and it's a powerful setup) 50! seconds to calculate the stuff. This is a lot for an open system. Any user could simply drown the server for every body by simply asking for that information 100 times a second.
This was the result:
by Oli on 15 February 2017 - 21:02
Explanation: Hardiman's formula only applies to 5 generation while Wright's requires the coefficient of every ancestor, which is why Hardiman's stops calculating after Generation 5
by Snowleaf on 16 February 2017 - 16:02
p.s. I tried to read up on Hardiman, but the math is quite complex. Do I get this right that with Hardiman you don't need to do a COI with more than five generations in order to get a "real" percentage of inbreeding? Even if, say, the common ancestors don't show up in the first five but only in generation nine or so? Does Hardiman take that into account? The Hardiman coefficient comes up with a much higher rate of inbreeding in your example, I added up the Wright percentages and the Hardiman's, and the first says only 0,93 as opposed to 7,01.
At any rate it seems as if the latter generations really don't seem to contribute much to the calculation of the COI. Which is strange, since C.S. Sharpe from Ashgi argues otherwise. According to her, seven generations is misleading in some cases.
by Oli on 16 February 2017 - 17:02
I will look into allowing users to use a higher number.
This page explains the difference really well: COEFFICIENT OF INBREEDING
Wright's Equation is haphazardly calculated to any number of generations, whereas Hardiman's Method is always calculated to five generations.
by Snowleaf on 16 February 2017 - 18:02
by susie on 16 February 2017 - 19:02
Oli's example may seem to show a lot of linebreeding but never forget: only in the "10. generation" there are 1024 dogs - 10 times the same dog out of 1024 f.e. is less than 1 percent - and those 10 breedings won't do us the favor to produce clones ... And thousands of different breeders won't select for the same traits...
by Snowleaf on 16 February 2017 - 22:02
"Aren’t genes several generations back too diluted to worry about?
The idea that genes get “diluted” across the generations is a common misconception. The genes don’t dilute, but every ancestor on average passes half of whatever genes it had to its offspring – one from each gene pair. Statistically, a dog will have half of a parent’s genes, a quarter of a grandparent’s, an eighth of a great-grandparent’s and so on. In actuality it may have more or less than that number but in most cases whatever came down will hover around those amounts. A dog five generations back statistically only passes on 3.13% of its genes. Chances are low that your dog has any particular one of that ancestor’s approximately 20 thousand genes; however he almost certainly has a few. After all, 3.13% of 20 thousand is just over 626 genes.
If a dog appears in a pedigree multiple times, then it is more likely that its genes will be passed on. If you have enough 3.13%’s from the same dog and maybe a 6.5% for an appearance on the 4th generation the percentage of genes contributed can add up even though that dog doesn’t show up on a 3-generation printed pedigree. This is a “percentage of ancestry” calculation. Such calculations do not take in to account the fact that different ancestors that are close kin are likely to be carrying many of the same genes so a line-bred pedigree is apt to have more transmission of the same genes down the generations than would be apparent if you only look at how many times a particular dog shows up."
You must be logged in to reply to posts