Real Coffee with Scott Adams - December 16, 2020


Episode 1221 Scott Adams： Using DNA to determine your COVID-19 Risk


Episode Stats

Length

11 minutes

Words per Minute

166.67604

Word Count

1,975

Sentence Count

129


Summary


Transcript

00:00:00.000 All right. Today, we have a very special single topic, and I'm here with Razeeb Khan,
00:00:07.700 who is a spokesperson and geneticist associated with TraitWell, and we're going to talk about
00:00:13.380 DNA and its potential to find out who has maybe worse risk with COVID-19. How are you, Razeeb?
00:00:21.700 I mean, I don't have COVID-19, so.
00:00:23.640 Yeah, you're already ahead of the game right there. So tell me if somebody wanted to check
00:00:28.840 their DNA, and we'll talk about the science behind it and the privacy and all that, but just start out
00:00:35.300 with what would they do if they wanted to know today, what would somebody do to find out if their
00:00:40.720 DNA suggests a higher risk for COVID-19? Yeah. So, you know, basically within the last six months,
00:00:48.540 there's been a lot of research on COVID-19, and people have looked at populations of people who
00:00:54.820 have been hospitalized, not hospitalized, they're infected, not infected, and tried to figure out
00:01:01.340 what the susceptibilities were, and there was a bunch of genetic hits that came up in the literature,
00:01:05.960 genome-wide associations. Some of your listeners have probably heard about those.
00:01:09.880 There's one on chromosome three that basically, if you have the susceptibility allele, which basically
00:01:16.280 means, okay, you got a marker that says you're more susceptible, you're about two times as likely
00:01:20.480 to be having severe COVID, which means hospitalized, you know, it's, it's, it's a, it's not just the
00:01:27.420 flu. Okay. Yeah. Now, help me with these statistics. Should I really care about doubling my risk? Because
00:01:35.280 you're doubling such a small number. Would I even know the difference? Depends on what your other risks
00:01:40.260 are, probably. Like, if you're, if you're, if you're 18, you're asking a very good question,
00:01:45.180 in terms of what are you talking about, Razeeb, right? If you are 72, let's say you can't retire
00:01:52.720 for some reason. So I'm, you know, obviously, I'm not a medical doctor, I'm not making recommendations,
00:01:57.120 but life is about trade-offs. I feel like our discourse forgets that. Sometimes you need to
00:02:02.880 work. We need to have people in the supermarkets, you know, we need to have people driving trucks,
00:02:07.140 we need to have people doing various things. But there are certain risks that go with that. And the
00:02:12.320 risks are not the same for everybody. If you have type 2 diabetes, if you have a heart condition,
00:02:16.720 if you're immunocompromised, we can go down through the list. A lot of these are associated
00:02:20.940 with genetics, right? So right now, there's a website that people can go to and enter not only
00:02:27.540 their DNA, if they have that file, but also their comorbidities. What is it? What is the website and
00:02:34.140 who accesses it and how? Yeah, it's covidforecaster.com. It's associated with Traitwell. And so if you go to
00:02:41.360 at Traitwell on Twitter, T-R-A-I-T-W-E-E-L-L. Basically, what we're doing there is we're
00:02:49.080 taking the information from your DNA and the information that you're giving based on, you
00:02:54.100 know, I mean, you could lie about your weight, that's up to you. But, you know, there's been a
00:02:59.400 lot of information of the comorbidities. And so that affects your risk calculus as well as well as
00:03:05.300 your genetics. So, you know, let's be entirely frank. If I had to know about your risk on chromosome
00:03:11.400 three, that particular position, as opposed to whether you're an 85-year-old with heart disease,
00:03:17.200 obviously, I'm going to look for an 85-year-old with heart disease, right? So I don't want to
00:03:22.680 dismiss environment or the, quote, non-genetic aspect. So how accurate should we assume that
00:03:32.640 this is? I mean, it's impossible to know yet, right? Yeah, we're still getting information. So I would
00:03:37.920 say, let me speak to the genetics part, because the environmental part is, you know, environment's
00:03:43.340 environment and depends on your test population. So the genetic part, there was a paper that came out
00:03:48.220 in New England Journal of Medicine using Spanish and Italian data in the spring, and Spanish and
00:03:55.440 Italian hospitals in Spain, hospitals in Italy. They found 5 to 10% of people in Western Europe had
00:04:01.120 this risk locus, had this risk marker, right? And this is severely associated on the order of about
00:04:08.140 like two, you know, odds ratio of like, you know, a factor of two against your just general risk.
00:04:14.120 They replicated that, 23andMe, the direct-to-consumer genomics company, they replicated
00:04:19.420 that result with a much larger American sample set. It's about the same number, like almost two,
00:04:25.880 right? And so if you're asking me personally, you know, one publication, even if it's peer-reviewed,
00:04:31.840 who knows? But the fact that they found it in another much larger population in the fall in
00:04:36.520 another country suggests to me that, yes, there are underlying genetic factors around that position,
00:04:41.920 and also a secondary effect, like a 25% chance increase risk on a chromosome 9. So, you know,
00:04:49.400 most of these characteristics have like all these little genetic effects. So for example, height or
00:04:54.240 intelligence. What we're here, what we're doing here is focusing on something that's kind of common,
00:04:59.760 so 5 to 10% is kind of common, and has a non-trivial effect.
00:05:04.520 Wow. Okay. And now the public has gotten this great lesson in science that non-scientists,
00:05:13.740 you know, hadn't had until COVID. And so we've trained the public that there's, you know,
00:05:19.340 the gold standard, you know, random controlled testing, and that can tell you something useful.
00:05:25.200 And if there are two of them, that's better. Where would you put this on the scale of 1 to 10,
00:05:30.700 where 10 is a randomized controlled, let's say, a reproducible study? So that's a 10. Where's your
00:05:38.740 reliability for this? Because this is a different kind of animal.
00:05:43.640 Yeah. You know, those are the gold standard. I put it like seven and a half because it's been in
00:05:48.760 large populations in different countries. And also, I mean, you know, this isn't like a casual
00:05:57.120 trait, like people really care, and they're tracking this in hospitals, right?
00:06:00.960 Right. So the odds are good that you're on to something. There's just always a caveat that,
00:06:07.160 you know, science changes, and what we know changes. So how do you deal with the privacy
00:06:14.660 element? I know enough about it that if I've taken a 23andMe or some genealogy kinds of sites where
00:06:22.520 you've already uploaded maybe your DNA, if you've done 23andMe, as I have, I can just download it
00:06:27.500 from my site. My app lets me do that. And then I can just upload it to your site. And I actually did
00:06:33.200 that. And it's, you know, minutes. I mean, there's no friction. There's no friction at all. Just
00:06:38.440 download and upload. Well, I mean, yeah, I mean, you know, obviously, everyone has to take that
00:06:44.120 seriously. You know, there's only, okay, let's be like, entirely frank, like, make reasonable,
00:06:50.160 rational calculations. I've talked to engineers. It seems like he's making a good faith effort at
00:06:55.760 that. And also, in terms of privacy, I mean, I think, honestly, the weakest position is not the
00:07:04.440 direct-to-consumer genomics companies. It's the hospitals. In terms of database breaks and stuff
00:07:11.160 like that, you know, this is just a pet peeve of mine. I don't know why hospitals don't get much
00:07:15.740 flack for this. Because a lot of these companies have had issues. But hospital databases are
00:07:24.360 notoriously creaky. So in terms of privacy, I think there are a lot of low-hanging fruit out
00:07:28.800 there. And I don't want to minimize it. But in the grand context, yeah.
00:07:32.400 But if you're comparing your risk of losing your DNA to, I don't know, the internet or something,
00:07:38.640 how controlled is the app? I mean, do you feel that you've got a good handle on that privacy-wise?
00:07:47.320 Yeah, I think so. You know, I think it's encrypted in a database that's, you know, not,
00:07:54.240 it's not like, I mean, SQL injections, all the standard things. You know, we did check for that.
00:08:00.500 Okay. Yeah, my thing on DNA and privacy is that I'm not going to try. I'm making my DNA. Anybody
00:08:08.500 wants my DNA, I'll make it public. I'll publish my whole DNA. But I don't recommend that for anybody
00:08:15.260 else. It's just a personal choice. I just don't think that should be a secret. Now, is it possible,
00:08:23.400 somebody asked me this question when they knew I was going to talk to you, could this sort of thing
00:08:28.100 be done in the future with a, maybe a little different technique to find out if somebody
00:08:32.880 would have a bad reaction to the vaccine? Yeah, but the data has to come in. So, you know,
00:08:39.340 unfortunately. Right. Not with any data that we have, but is there, the process could identify
00:08:46.840 that there's a DNA correlation? Yeah. Yeah. Would you suspect that there is? Yes. I'm going to be
00:08:55.800 honest. Like, most things that are biological have some heritable, you know? A lot of times,
00:09:03.260 we don't care. Let me tell you a story. I had a friend who died induction into the Marines from
00:09:11.680 the anthrax vaccine. This happens every year, but it's a risk that we take, right? Don't talk about it
00:09:18.760 very much. I know about it because friend, right? Probably, if I had to bet, he had some very rare
00:09:25.640 mutation that doesn't have any problem otherwise. Right. You can't predict that. No. So anybody who's
00:09:33.020 joining us now, there's already a site where you can put in your DNA and other comorbidities.
00:09:39.960 It's called covidforecaster.com. It's exactly like it sounds. Spell it the way it sounds. And
00:09:48.260 do you have any other warnings or caveats about using the system? My take, based on what I've seen,
00:09:58.800 is that I would add it to the many other risk management variables that you're weighing when
00:10:06.780 you're deciding everything from, you know, how much should I fight to be first in line to get
00:10:11.500 that vaccine to how much should I, you know, go crazy on social distancing versus, you know,
00:10:17.180 being a normal, responsible person. But I wouldn't say it's the, you would agree that you can't say
00:10:23.320 it's a hundred percent anything for any one person. It's just one more variable to add to a lot of
00:10:29.080 variables. Yeah. These sorts of things are updating and changing over time based on the information we
00:10:33.520 have, right? So you update your priors based on what you have. We know so much more in six months
00:10:39.360 in a good way and in a bad way about this sort of issue. So that's one thing I would say. It's
00:10:45.300 always an evolving thing. Your DNA doesn't change, but the information we have about it does.
00:10:49.960 So this is great. I just wanted to keep this short and tight, let everybody know that there's a way
00:10:56.720 they can upload their DNA. Now that does the company commit that it doesn't, it doesn't sell
00:11:02.400 the DNA. It doesn't release the DNA standard standard. So you'd have the same risk that you
00:11:08.960 have where it is, which is if it's already a 23 and me, it's already in somebody's database that this
00:11:15.700 is, this is the reason that I don't obsess about protecting my DNA. I just don't think it's the thing,
00:11:20.560 you know, in the same way that I don't think I would ever have privacy in digital communication.
00:11:27.720 I mean, I want it, but I have, but I live my life like it's not really a thing. So it's the
00:11:33.020 same standard I use. So it's covidforecaster.com. Razib Khan, thank you very much for sharing this with
00:11:40.960 us. And it's a one more tool. We'll add it, we'll add it to the good news that's coming out on COVID
00:11:46.040 because we got enough bad news. All right. And have a great day.