What can genetic testing tell you?

A while back I took a 23andMe genetic test that looks at over 600,000 different spots on my DNA. The last few blogs I have been going over my genetic test results with an eye on how useful they are. And how well the results are explained.

Last blog I wrote about how current genetic tests aren’t that great at predicting your risk for common, complicated diseases like diabetes or Alzheimer’s. This time I thought I’d focus on what today’s genetic tests can be very good at and whether or not 23andMe does a good job with these.

Current genetic tests are very good at predicting your risk for rare, simple genetic diseases like cystic fibrosis (CF) or Huntington’s disease (HD). And at predicting the chances that your kids will get these diseases too.

Genetic tests for these diseases work because most of them are caused by a single gene gone awry. Testing for a single gene is relatively easy.

For example, most cases of CF happen because of known differences in the CFTR gene. A genetic test can look for these differences and tell you if you and/or your spouse have any of them. If you both do, they can also give you a pretty good idea about the chances that your kids will get them too.

Of course, we don’t know all of the differences in the CFTR gene that can cause CF. And some differences only cause CF some of the time. And there are people with everyday, run-of-the-mill CFTR genes who get CF because of differences in different genes.

Still, as genetic tests go, these are pretty good. If a test comes up with a known CFTR difference that causes CF, then you have a pretty good idea of what your chances for developing CF are. If your spouse gets tested too, then your kids’ chances can be determined as well.

So how does 23andMe do? OK, I guess…

First off, they look at eight of these sorts of diseases under a category called Carrier Status. The diseases they look at are shown in this image:

For me, the first big result is that I am a carrier for a variant that can lead to hemochromatosis. This isn’t surprising since 1 in 8-12 people of Northern European descent in the U.S. are too, but it is definitely something to watch out for. It may be important for my wife to be checked too so we can make sure none of our kids got two copies. (Luckily hemochromatosis is easily treated by giving blood on a regular basis.)

Some of the other results are less illuminating. For example, I do not carry the CF difference they test for (delta F508). This is of course great news. Unfortunately, this variant only accounts for about half of the CF cases out there. Which means I could be a carrier for CF, just not a carrier of the most common variant that they happen to test for.

The same thing goes for most if not all of the other carrier status diseases (sickle cell anemia is an exception). Some like BRCA (breast cancer) are as poorly covered as CF while others like Bloom’s disease cover a larger percentage of cases.

23andMe is pretty upfront about the limitations of their testing once you dig a bit into the results. But still, if they’re going to look at 600,000 different parts of my DNA, you’d think they could add a few more to give me a stronger answer about whether or not I am a CF carrier.

Could a genetic test have told me I was at a higher risk for developing type 2 diabetes? Image source: aldenchadwickThis sounds contrived but it isn't.  A couple of weeks ago I was diagnosed with metabolic syndrome.  Right when I am in the middle of talking about genetic testing!

Metabolic syndrome isn’t quite as scary as it sounds.  Basically I am on my way to type 2 diabetes.  But if I eat better and get off the couch, I should stave off the disease and get better.

My question, naturally, is whether or not a genetic test could have told me I was at a higher risk for developing type 2 diabetes.  And whether I would have done anything with that result.

As you know if you’ve been following my blog, I took a 23andMe genetic test and have been writing about it since.  The image below shows what the front page of my clinical report looks like (click to enlarge):

According to the DNA checked in this test, I am in the average risk range for type 2 diabetes.  This doesn’t really seem to line up with my reality.  But I might not expect it to since these genetic tests are so limited right now.

This kind of test can be informative with the yes answer—yes I carry a certain version of a gene that might lead to a disease.  But the no answer isn’t that useful.  It doesn’t mean that they've looked at all the possible genetic differences that can lead to a disease and I don’t have any of them.  Basically it means that they didn’t find the specific genetic difference they were looking for.

Now I wouldn’t necessarily have predicted that any genetic test available right now could tell me a lot more than that.  Type 2 diabetes is too complicated for that and a whole lot more research will need to be done to get a genetic test useful to lots of people.

But still, this is probably what people are looking for with these sorts of genetic tests.  Will I get cancer, type 2 diabetes, Alzheimer’s, Parkinson’s, etc.?  For most of these cases, the tests can tell you a lot about rare forms of these diseases but little about the more common forms.

So the no answer didn’t really help me much.  Here I am on my way to being a diabetic and the test said I was at average risk.  Of course, I suppose I didn’t even need to take a test… all four of my grandparents came down with type 2 diabetes.  Like lots of these complex diseases, family history is the best predictor.

The second part of my question is a hypothetical one.  Let’s say they had a perfect genetic test that said that I was at an increased risk for type 2 diabetes.  Would it have changed my behavior?  I’m not sure but probably not.

I certainly wouldn’t have changed any of my behaviors when I was young.  I was invincible, remember?

Now that I’m a bit older, such a test might have influenced my behavior a bit.  I already knew about my risk because of my grandparents but my thought has always been that maybe I got lucky and didn’t inherit their tendencies towards diabetes.  But if they were tested and we shared the same genetic differences that led to type 2 diabetes, then I might be worried enough to change what I was doing.

Most likely though, my behavior modification wouldn’t be perfect.  What I’d probably do is keep watching TV and eating Twinkies but get my blood sugar tested more often.  Once I was headed for diabetes, then I’d modify my behavior and keep it at bay.  (I’m sure doctors scream into their pillows at night because of patients like me.)

This is different than some people’s reactions to other genetic tests.  For example, some women who find out they have the version of BRCA1 that greatly increases their chances of breast and ovarian cancer have a double mastectomy and/or a hysterectomy before there are any signs of cancer.

I might react much more strongly with a valid cancer genetic test.  Cancer is scary, nasty and not really reversible.  Type 2 diabetes is different.  You can start down the road, modify your behavior and then nip it in the bud.  Carpe diem and then pay the piper.

Are they related to me? I still don't know…When last I left you, I was searching for my great-great grandmother’s DNA in my own DNA.  Remember, legend has it she was Cherokee and I wanted to confirm the legend with a genetic test from a company called 23andMe.

In my last blog post, I showed how the two most powerful ancestry tests, mitochondrial DNA (mtDNA) and Y chromosome, were useless to me in my hunt. Now I want look at the rest of my DNA.  So here we go!

The Y chromosome and mtDNA are a small fraction of my DNA—something like 0.8% of the total DNA in one of my cells.  But they are incredibly useful because they change very little from generation to generation.  The mtDNA I got from my mom is probably exactly like hers.  Same with most of the Y I got from my dad.

The other 99.2% of my DNA is a lot trickier to look at from an ancestry perspective because it has changed a lot from generation to generation over time.  For example, the chromosomes I inherited from my parents are not the same as the ones they have.  I got a mix of their chromosomes

For example, my mom had two copies of chromosome 1 (and two copies of her other 22 chromosomes too).   As you know, she passed one chromosome 1 to me (my dad gave me my other one).  But, through a process called recombination, her two copies of chromosome 1 swapped DNA so that I got a hybrid of her two copies.  I inherited a unique chromosome never before seen.

This is all well and good from a survival of the species point of view, but it is a problem for ancestry testing.  Imagine that instead of my mom, we look at my Cherokee great-great grandmother.  She has just had a child who inherited a mix of her chromosome 1’s.  This chromosome will look Native American and the child would appear half Native American.

Actually, the test isn’t perfect yet and so there isn’t yet a “Native American” set per se.  Instead, here is how 23andMe describes Native American DNA in their tests:

“…people who identify themselves as Native American exhibit fairly consistent Ancestry Painting proportions of about 75% Asian and 25% European, plus or minus 10%.”

This means the chromosomes the child got from his or her mom won’t look Native American but instead will look 75% Asian and 25% European.  (See a realted post of mine elsewhere for why it looks like this.) Now imagine that this half Native American child grows up and has my grandfather as his or her son.

My grandpa will inherit a mix of his parents’ DNA too.  In this case the Native American DNA will mix with the European DNA to create a hybrid.  On average, you would now see something along the lines of 37.5% Asian (this is a simplification but it gets us into the ballpark of the number we might expect).

Each generation would see, on average, a continued dilution of this Asian part.  My dad would have 18% Asian, I would have 9%, etc.  Here are my ancestry results (click the image to enlarge):

Not a hint of Asian.  Looks like my great-great grandma wasn't Cherokee.  Or was she?

There are lots of ways she could still be Cherokee.  First off, I don’t know how solid the 75% number is for all Native Americans.  I don’t know how many Native Americans are in their database.  I also don’t know how much variation there will be tribe to tribe.

Secondly, you may have noticed that I was very careful to always say, “on average.”  This is because the recombinations don’t have to be a 50-50 swap.  It is true that if you look at a large number of recombination events, the average will be 50%.  But individual recombination events can be biased towards one or more chromosomes.  Occasionally you’ll get mostly one chromosome and sometimes mostly the other.

Sort of like flipping a coin—do it enough and you’ll get pretty close to half heads and half tails.  But if you flip a coin twice, you might get one head and one tail.  And you might not.  Half the time you’ll get two heads or two tails.

This is less a problem than you might think with our chromosomes since the recombination is spread over 23 pairs with each pair being independent of the others.  But it can still throw a monkey wrench into the works.  23andMe actually has a nice chart that hints at this by giving the most likely range of possibilities.  Unfortunately, this chart didn’t come up with my results and I had to stumble on it while I was playing around.

Using the chart, I can see that the bottom end of my expected results in 0.24% “Native American” (if I am reading the chart correctly).  That is pretty low and it seems like a pretty minor mistaken assumption at the beginning might knock this down to zero.

So where am I after this?  Still in the dark.  This is actually how many genetic tests end up.

The positive result tells you a lot.  Had there been Native American DNA, that would have been a slam dunk.  (This isn’t always the case with genetic tests but it would be here.)  But there wasn’t.  Which means, given that I was on the edge of detection, that she may or may not have been Cherokee.

Now, this isn’t 23andMe’s fault.  The test itself couldn’t be conclusive given how far back we need to go and the DNA tests that 23andMe offers.  In fact, 23andMe does an excellent job of presenting the data.  There are pretty chromosome paintings, graphs superimposed on world maps, etc.  All very nice.

I am still worried that the explanations that go along with these images assume an awful lot of knowledge that most people might not have.  Without that knowledge, it can be hard to assess the significance of a certain result.  Next blog that’ll become even more important as I tackle health conditions.

Genetic tests often don’t give as much information as you might think.In a previous blog I talked about getting my DNA tested with 23andMe.  Well, I got the email the other day saying that my results were ready.  So I logged on and up popped this screen pictured to the left.

All sorts of goodies to try out!  I feel like a kid at Christmas.

The first thing I thought I’d do is check out my ancestry.  My grandfather’s grandmother was supposedly Native American and so I wanted to find out if I could see that in my DNA.  (This relates to my supposed relationship with the outlaw Sam Starr but that is a different story.)

23andMe has this Native American testing app in their 23andMe Labs section.  I clicked on my data and up popped this result:

Recent Native American ancestry is unlikely

Has it all been lies?  My great, great grandma wasn’t Native American?  Not so fast…

A “no” answer on a genetics test doesn’t necessarily tell you a lot.  (And sometimes, the “yes” answer isn’t so helpful either!)   Now as a geneticist, I know the drawbacks of ancestry tests like these.  What I wanted to see was if 23andMe did a good job of explaining them.

I first checked out my mitochondrial DNA (mtDNA) and my Y chromosome data.  These DNA don’t change a lot from generation to generation and so are really good at tracing ancestry many generations back.  Their downside for me is how they are passed down.

The Y chromosome passes from father to sons.  My great, great grandma didn’t have a Y to pass on so of course my Y chromosome data wouldn’t show that she was Native American.

mtDNA passes from mom to her children.  At first this sounds promising since we are talking about my great, great grandma until we realize that I am related to this woman through my grandfather.  His mtDNA died with him (except for his female relatives and their descendants) so that is lost to me as well.

Here is what 23andMe has written under interpretation of my mtDNA and Y chromosome results:

This mitochondrial DNA haplogroup is inconsistent with Native American ancestry along the maternal (mother's mother's mother's …) line.

This Y chromosome haplogroup is inconsistent with Native American ancestry along the paternal (father's father's father's …) line.

I suppose this says what I just said but I am not sure how many people would really appreciate the limitations of mtDNA and Y chromosome data from this explanation.  There wasn’t a link to a more explicit discussion of the limitations of this sort of testing and there wasn’t anything I could see from a quick glance at the ancestry part of the site either.  An explicit explanation would be good or maybe a figure like this one:

To me, this drives home the point that there is a whole lot of missing ancestry.  It might help if they had some sort of family tree app where you could indicate as much as you know about family relationships.  Once you’ve inputted the data, it would spit out what tests results would be useful to look at.

So the mtDNA and Y chromosome test results are of little use to me in this quest.  (And of little use to me in general as it confirms my pasty whiteness.)  Next blog I’ll deal with the rest of my DNA and what that can and can’t tell me about my great, great grandma.

Hopefully this DNA analysis data will be better at telling my future than tea leaves or goat entrails.Well, I have finally decided to do it.  I have ponied up the money and signed up for 23andMe's DNA test.

This is a test that will look at over 500,000 different spots on my DNA.  From the results I'll be able to learn about my future health and my past ancestry.  Well, as much as I can learn given the current state of genetic knowledge.

And there's the rub.  I have held off on doing this for quite awhile because I am just not sure how useful it will be.

Most of the DNA studies on the big diseases like schizophrenia, autism, diabetes, heart disease, etc. have not been that conclusive.  They tend to find bits of DNA that have a very small effect on risk.

Undoubtedly as more studies are done, we'll find lots of bits of DNA like this and we'll be able to figure out our risk more accurately by adding them all up.  But we're not there yet.  In fact we're probably years away from being able to do this.

I have also been a bit squeamish about sending my DNA to a company.  Yes, I know they'll be careful but still…it's my DNA.  You can't get any more personal than that!  I would hate for someone to get that information and use it against me (think insurance agent).

So why did I finally decide to go through with it?  One reason is that I get a lot of questions from people at the Ask a Geneticist site about how useful or good the test is.  Right now I have to tell them I don't know.  I'd like to be more helpful than that.

I also think that it will be fascinating to see all of my bits of DNA.  This is the stuff that is a big part of making me who I am.  It will be so cool to look into that crystal ball even if the future I see is a bit murky.

Of course as a big old science geek I'll be interested in that stuff…it's my bread and butter!  What I also want to do is try to imagine what the test is like for someone who doesn't go all gaga for genetics.  How is it for people who aren't necessarily mesmerized by the beauty of DNA and instead are mostly interested in diseases, traits, and ancestry?

I guess I'll find out soon.  I sent my spit in last week.  I'll keep you updated in future posts.