A visual representation of the KKA Kai Ken population (so far)

I've been working on mapping the Kai Ken Aigokai (Kai Ken) population. All I have is what is in the registry that I have built based on our imports plus some other dogs that I've gotten info on. I want a visual way to look at the entire population to help us find potential bottlenecks. This will help all of us better understand just how close the population is and what lines to avoid due to these bottlenecks.

Here is my first go at it... I still need to clean it up some. The blue dots are Kai Ken we have in our kennel, the red dots represent Kai we have produced.

NA-KKA

Click here to view the full resolution version.

As you can see, even with all the work we have done to import Kai in groups that are as unrelated to the last import group as possible, there is still a very close relation between all of them.

----
«1

Comments

  • Curious on what do you call this type of mapping? What program generates this?
  • This doesn't include the Ayu x Akashi litter?
  • @ayk - I built this using Cytoscape. The data was output from the registry app in a format that could be imported into Cytoscape. The mapping above is a force-direct organic mapping. The distance between nodes is supposed to represent their relative relationship to all the other nodes (Kai).

    No, this doesn't include Ayu's litter.

    I plan to make a version that shows only the KKA Kai in North America, which wouldn't include any of the ancestors that are not directly related to all the dogs already here (this would show how closely related all the KKA imports are to each other).

    ----
  • edited April 2012
    I am curious too. :)

    I have a database of about 17 000 entries, and I wonder how the result would look.

    Edit: Nevermind, Brad answered. Thanks.
  • edited April 2012
    Here is another version that shows "network-neighbor connectivity" ... basically, how each node is connected to it's closest relative. This helps to show which dogs are more connected to the group as a whole (which implies that their line is used more in the breeding population)

    NA-KKA-network-neighbor-connect

    Click here to view the full resolution version.

    ----
  • @souggy - You should check out Cytoscape! There is a bit of a learning curve but it's super cool once you figure it out. I'm totally geeking out over it right now.
  • edited April 2012
    @*JackBurton* this is neat you should check it out

    Bah apparently I can't tag Sean. :(
  • edited April 2012
    I see Haru and Nio!
    :)
    (in the 1st one. the second one is a ball of string)

    art brain checks in!
    We have a book upstairs: "Visual Representation of Quantitative Analysis" (I think) all about these kinds of graphs and webs and so forth and various ways to show data visually. Check out Edward Tufte if this sort of thing rocks your boat.
  • Edward Tufte is the shit!
  • Is the Cytoscape force-direct mapping supposed to be represented in 3D?

    Just wondering because littermates aren't exactly the same distance between their parents.

  • Oooh! Nifty! The analyst in me really wants to see the data for these charts and play around with it... :)
  • @poeticdragon - Yea, that's annoying.

    @ayk - Yes, it's 3D... and I moved some of the litter mates so we could better see them since in the graph they were next to each other on the z-axis. The app doesn't have a great way to out put a 3D file that would be usable on the web.

    ----

    This is all very much a work in progress, but I thought some of you might enjoy seeing it.

    Here is a positive side effect of doing this that I had not considered. Reviewing the graph I noticed this, which seemed pretty odd (nodes 66 & 72)...

    issue

    So I looked it up. Ah, yea, Kai Kokuou #32135 produced his sire (link).... huh??? Opps! I guess I messed up entering someone's pedigree...

    So, it's already helped me find a mistake in the DB.

    ----
  • Is this software free? And how many dogs did you have entered?
  • So, this will help us to find lesser related individuals to make pairings with greater diversity?
  • @poeticdragon - Yes, it's free. JAVA based. I have around 280 KKA Kai in the DB.

    @lindsayt - Yes, that is the hope. It will be more revealing as we get more data (more Kai)... We will start to see popular studs and other types of bottle necks.

    I'll make a new map to point out some of the current potential bottle necks I'm concerned with. I'll post it later.

    ----
  • @brada1878 what format is the input data in?
  • Thanks! And nice work :)
  • edited April 2012
    This is amazing! I just zoned out on the graph for a half an hour! This is definitely a first. You are THE Kai geek Brad! lol
  • That's amazing. I can't wait to see more? Curious if Kitora is on there... I'm looking but might not be familiar enough yet w/ different litters etc....???
  • @poeticdragon - The Kai reg DB is MySQL... I exported it as CSV and then imported it into the app.

    @tjbart17 - I am a Kai geek, huh? lol

    @CarabooA - Nami and Kitora aren't in the DB yet as I am waiting on clarification of #26 in their pedigree. I need that Kai's reg # before I enter them (so I don't end up with duplicates in the DB).

    ----
  • @brada1878 - Ok, good....So, I'm not that much of a goober for not being able to tell where she and Nami are...lol Well, I'm probably still a goober but - whatever! :oD
  • woah. nice. Who's the outlier that's like 8 degrees away from any of the others? My guess is Tyson
  • @caraboo I think I'm the goober for not really understanding how to read it at all. I thought, oh, cool, but I don't get it, and my brain did the "ok now I'm shutting down because this could be math related" thing it does! (normally I'd be embarrassed to admit this, but hey, studying anatomy and patho/phys has just taken away any sort of shame I have about not getting stuff).

    More seriously, I'm getting that the blue middle dot must be Yamabushi kennels? And then the other dots are other Kai Ken and their relation to other dogs? Am I right on that? So in that way it is working like a different sort of family tree type representation, with location included?

    And the second one, the circular one, is too mysterious for me to even guess at.
  • @shibamistress--don't worry, this is how I felt when I first saw these "balls and sticks" too. But once you get the hang of "how to read it" and understand the working logic behind, it is a very powerful tool in population genetics as well as other subjects. I forgot what program I used to this before (5 years ago), but we use this to see how closely orthologs of genes are related at different levels amongst various species.

    @brada1878--this is way cool, thanks for taking the time to put this together, it is quite educational for all of us. I was actually planning to put something like this together for all the shikoku eventually but THESIS comes first!!!
  • @shibamistress - I'ts a chart showing relatedness of registered Kais. So, the more "in the mix" the dot is, the more related they are to the rest of the population. That's my guess anyway. The blue ones are all Yamabushi. Red are their offspring. My guess is the middle one is Nio, to the left Kumi, to the top right Haru. outside of that i could spend hour speculating. Though there are numbers attached to each node, so there's probably a much easier way to find out. :)
  • This is super cool! Thanks for sharing this Brad.
  • @Yandharr - The one way off to the left is Hana... Hint's why we imported her, she is relatively unrelated, how ever she is very related if you consider how she is related. Which I know makes no sense the way I wrote that... but basically, she is related to ALL of our lines in some way, where some of the other imports, like Sachi, are only related to one line (tho much closer to that line than Hana).
  • edited April 2012
    Also, the numbers in the graph are each node's ID in the DB.

    So, if you want to see who is who you can use the reg app by replacing "[id]" with one of the number in the graph of this URL: http://www.yamabushikennel.org/apps/pedigree/index.php?r=dog/view&id=[id]

    For example, here is Hana's (id 330), URL: http://www.yamabushikennel.org/apps/pedigree/index.php?r=dog/view&id=330

    ----
  • Do you have a version with the image big enough to read the numbers? I tried the largest on flickr but still couldn't see them.
Sign In or Register to comment.