I sent my dog's DNA to
three different DNA testing companies and the results all
came back a little different. Let's talk about why. This video is
sponsored by Datacamp. We're going to be talking about
analyzing genetic data today. And that's all done
with bioinformatics. In grad school, I used
Python and R all the time. But when I got
there, I didn't even know how to use the terminal. So I learned it via lot of
Google and help from friends. And this is where
something like Datacamp would have been awesome. Datacamp is an online
learning platform that makes it easy to build
data analytical skills. You can learn at your own
pace, with interactive courses and hands-on exercises. And it really is for all
skill levels from Excel 101 to advanced courses. I have encoded in a
couple of years now. And a refresher would be great. So I took a Python assessment to
find out where I should start. And it placed me right
at the edge of novice and intermediate. And that totally feels right. So I'm going to go back
and relearn some Python that I have forgotten. Datacamp doesn't require
any special software. You can do it all right in
your browser, which is awesome. So it's a really easy
way to learn or refresh some coding and
analytical skills. If you want to do things
like analyze genetic data, learning Python and R
super, super helpful. Subscriptions start at $25
per month for unlimited access to courses. But you can use my
link in the description down below to check
out the first chapter of any course totally free. This is Penny. We adopted her in May of 2020. And she is just an
absolute sweetheart. Her favorite things in the
world are naps and chicken and more naps, especially
naps in the sunshine. She is the sweetest dog we
could ever have hoped for. She is so sweet and all
she wants to do is snuggle. And we love her
so much, so much. We got her from her rescue. And they told us that she
was a chihuahua/corgi mix. And chihuahua I can see,
but corgi, not so much. So I sent her DNA off it to
get some doggie DNA ancestry testing done. And I want to be clear, we
would love her just the same if the results
came back and said that she was happy chihuahua
had Loch Ness Monster. But I am a geneticist
and I couldn't resist. So I also didn't just send
her DNA to one company. I sent it to three. You can get down or over here. OK, you sit your
butt right there? You can sit your
butt right there. The way each company analyzes
the DNA is a little different. And to understand
why, we've got to talk a bit about the structure of DNA
and why there really isn't such a thing as a chihuahua gene. A chihuahua, a golden
retriever, and a German shepherd are all dogs. Even though they all look
super different, genetically, they are 99.9% identical. And this makes sense. They are mostly the same. They pretty much all have
limbs and brains and ears and fur and tails and toe
beans and "boopable" snoots. So they all have genes for
making those different parts. And usually, it's not just one
gene that makes up a trait, but many. There isn't just
one eye color gene, but many genes that all
contribute to eye color. And each of those genes can
come in different versions or alleles. For example, there
are a number of genes involved in coat length. And it's not that there
is a gene for short hair and a gene for long hair, but
rather, different versions of a hair length gene that
can predispose your dog to having short or long hair. And the difference
between these versions is usually really small. For example, there's
a position in a gene called FGF5, where a
Tbase is associated with having long hair. And a C-base is associated
with having short hair. And remember, that
like you, dogs have two copies of every
gene, one from each parent. So dogs that have a TT at
that position, typically have long fur. And dogs with a CC or CT
typically have short fur. So you can look at that
position in your dog's DNA and make a prediction about
what their traits will be. So when we think about the
whole dog genome, 99.9% of it is the same. But it's studded with
these small locations that differ between dogs. Some of them are SNPs, Single
Nucleotide Polymorphisms. And others are indels,
short insertions or deletions of a few bases. And there are also
small regions of repeats that can vary between dogs. But most of the DNA
is exactly the same. And really, it's just
about 0.1% of the locations in the 2.8 billion
base pairs long genome that give us all of the
different kinds of dogs in the world. So back to the tests. If 99.9% of the DNA
between dogs is the same, you don't really need to
look at that 99.9% of the DNA to learn something
about your dog. And the dog genome is over
2.8 billion base pairs long. So it can be time consuming
and expensive to look at parts that aren't going
to tell you much. So two of the companies,
Embark and Wisdom Panel just look at that other 0.1%. They do this using something
called a microarray. This is a chip covered
in tiny pieces of DNA that match to those
0.1% of locations, about 100,000 to
200,000 locations depending on the company. So if you add a dog
DNA sample to this chip and see what binds, you can
tell if your dog has an A, T, C, or G at each of these
specified locations. On the other hand,
Darwin's Dogs that does whole genome sequencing. This means that they read every
single base pair of your dog's DNA on a DNA sequencer. This captures those many
little regions of variation. But it also allows them to look
for new unidentified locations of interest as they
sequence more and more dogs and learn more about
the dog genome. So how can this tell me if
Penny is a chihuahua or a corgi or an Alaskan malamute? As you can maybe already
guess, since I told you that dogs are mostly
genetically the same, there aren't chihuahua genes or
husky genes or terrier genes. But there are variants or
a alleles of those genes and many genetic markers
along the genome that show up in some breeds
more than others. There are also
patterns of markers that show up more in
one breed than another. So before they can
analyze our pet's DNA, these companies look at the
DNA of lots and lots of dogs. And they figure out which
SNPs or alleles or patterns show up more in one
breed than another. So they look at all
the chihuahuas together and they say, OK, they're most
likely to have DNA that looks like this. Then they do the same thing for
the Huskies and the Dalmatians and the German shepherds and
the terriers and on and on. Then when they
sequence your dog, they can compare the patterns
of DNA in your dog's DNA to the types of dogs in
their reference panel and say, OK, at this
location, Penny's DNA looks most like the DNA
found in a chihuahua. And at this location, it
looks most like the DNA found in a terrier. And they do this
for the whole genome and they calculate percentages. So how did Penny's DNA look? OK, we're going
to view Penny's-- oh, I was wrong. 37% chihuahua, 25%
Cocker Spaniel, 14% American Staffordshire
terrier, interesting. Well, now we get to see what
the other two come back as. Oh, they sent us a video. Interesting. So as I suspected, the
results aren't the same. And I can't wait to
get the third one back. That's going to be fun. And according to
Darwin's Dogs, Penny is 22.9% Cocker
Spaniel, 20% chihuahua, and a whole bunch of
you know, other mixes. So all three of Penny's
ancestry reports came back and said that
her major components were a chihuahua and a Cocker
Spaniel, which was surprising. But then, all of the
other percentages were pretty different. Two of them had American
Staffordshire terrier. One of them had
poodle and Pomeranian. And there was
dachshund and pit bull. So the main components of her
results were all the same. But they weren't identical. And why not? They all sequence the same DNA. It's because each company is
using a different reference population. Imagine that the
company only sequenced two types of dogs, Dalmatians
and German shepherds. When I sent in
Penny's sample, they would have only had those two
types of dogs to compare it to. And so they would match her DNA
to the closest breed possible, or tell me that they
couldn't find a good match. If they then added chihuahuas
to their reference panel, now, they would have a
much better match and be able to update
her percentages with much more confidence. So because no company has
sequenced every single dog in the world, they
have to rely on trying to build the best, most diverse
reference panel possible. But each company's panel is
going to be a little different. So they might have
slightly different results when trying to assign breeds
to sections of Penny's DNA. And as the reference panel
grows and they sequence more and more dogs, their
confidence in their matches will get even better. That's why you might get an
update on your dog's breed mix. It's not because your
dog's DNA has changed. But rather, because
the reference panel it's being compared
to has changed. And if you've had
direct-to-consumer testing like 23andMe or
Ancestry DNA done, you might be wondering if
the same thing is true for us humans too. And the answer is, yeah. You might get updated ancestry
percentages every now and then. And it's not because
your DNA has changed, but because the reference
panel has changed. And the pattern matching
works in the same way too. My DNA test came back that I was
45% French and 25% Lithuanian. But that's not because
there are French genes and Lithuanian genes. It just means that 45%
of the markers in my DNA looked most like the
pattern of markers found in people with long
family histories in France. And 25% looked most like the
pattern of markers in people with the long family
histories in Lithuania. And again, these markers
make up a tiny, tiny fraction of the genome. You and me are 99.9%
identical at the DNA level, just like Penny and
Pandora and Java are. This isn't a review
of these companies. I'm not ranking them. I think they all did a good job. If you want to
test your dog's DNA to find out what breeds
it is, you really only need to do it once. And it's kind of
just your preference on what options or other
features these services offer that will make you
choose one or the other. Come on up. We got to say thank you. Come here. Yeah, good girl. OK, good girl. Thank you to Datacamp for
sponsoring this video. And thank you to
you for watching. As always, a huge shout
out to my Patreon patrons, who help me to make all
kinds of different videos about genetics. And also, a huge thank you
to everybody on Twitter who sent in pictures
of your pups that I could use in this video. I mean, I know people like
sharing pictures of their dogs. But I can't tell
you how much joy there is in just
getting a notification list full of dog photos. I loved it, even
if Penny didn't. And as always, remember to
go forth and do science. [MUSIC PLAYING]
