Our first question upon hearing that someone has a new baby is usually "Is it a boy or a girl?" But our first question upon hearing that someone has gotten a new puppy is more often "What breed is it?" Breed is at the heart of how we perceive dogs. It affects many of our expectations of them - energy level, intelligence, friendliness - for better or for worse. With mutts, however, our urge to make breed-based assumptions can be stymied by the lack of a known breed to which to attach those assumptions. And so when you have a mutt (as I do), you learn to play the "what is it?" game. Some people guess, some people make up clever names (I have a Golden Collie, or sometimes a Border Retriever) - and some people have their dog genetically tested to determine its ancestry. A new project, MuttMix , provides you with photos, video, and behavioral tidbits for mutts and lets you play the "What is it?" game to your heart's content, for the benefit of science. Karlsson Lab , of the Broad Institute of MIT and Harvard, has teamed up with the International Association of Animal Behavior Consultants (IAABC) to try to find out how good people really are at guessing the breeds mixed up in the American mutt. We won't release the answers immediately, because we want to be sure that no one has privileged information when they make their guesses, but after the project closes on June 16, we will release the breed mixes publicly (and start writing the journal article!). The project opens today, so go start guessing!
The breed calling algorithm behind MuttMix was designed by Dr. Linda Boettger as a postdoc at the Broad. Linda earned her PhD studying human genomics, and after graduating thought it would be fun to apply some of those same approaches to dogs. She explains, “Dogs are really the ultimate model system because they are the most morphologically diverse species of mammal and have been subject to intense selection pressures for both appearance and behavior. Plus, Elinor [the head of Karlsson Lab] told me that I could find out the breed ancestry of my dog, and all I had to do was write a ton of code!” Linda adopted her dog Skyler from a shelter, which told her Skyler was a Labrador/pit bull mix, but Linda was intrigued to find out that Skyler is actually a mix of many breeds, primarily Dalmatian and Rottweiler. (Sadly, Skyler passed away several months ago, but she was immortalized on the 2017 Karlsson Lab holiday card). To test her algorithm, Linda needed mutts with highly mixed-up ancestries, but she also needed to know for sure what those ancestries were - a situation hard to come by in the real world. So she turned to software to simulate crosses using real purebred dog genomes to see what their puppies’ genomes might look like. She continued these simulations for many generations until arriving at mutts with ancestry from many breeds, similar to the real dogs involved in the MuttMix project. Linda’s MuttMix breed calling algorithm "paints" the parts of each chromosome that have originated from a particular breed. In the case of Lucky, a mutt who lives with one of our lab members, each chromosome traces fragments to a variety of different breeds. By "painting" each chromosome fragment with its relevant breed heritage, we can see exactly where Lucky gets the traits that make up his unique look - drop ears from a toy poodle ancestor, a manly beard (or "furnishings") from a Lhasa Apso. We can even predict that Lucky would do well in a high altitude environment due to a mutation he got from his altitude-adapted Lhasa ancestors. Our citizen scientists who help us out with MuttMix may prove us wrong, but we're hoping to show that it's actually quite difficult to guess a dog's breed based on looking at the dog, even with some video and behavioral descriptions to help them out. We all know at some level that every dog is an individual, but sometimes we forget that when we remember how gregarious the last Golden Retriever we met was, and see how much the yellow dog in front of us resembles that one. Even within a breed, canine personalities vary widely. Through MuttMix, we hope to be able to provide some solid scientific evidence that you can't judge a dog by its fur color. But maybe you'll prove us wrong, so please, go to MuttMix, take a look at the wide array of cute mutts we have there for you, and give us your best guesses! Guest post by Jessica Hekman, DVM, PhD, a postdoc at the Broad Institute of MIT and Harvard, where she helps out with cool projects like MuttMix in addition to focusing on her own research, The Working Dog Project. Jessica has a mutt, Jenny, whose DNA is on the sequencer right now, and she can’t wait to see what Linda’s breed calling algorithm will have to say about this one. @dogzombieblog / facebook.com/dogzombieblog