>>Uh our next uh our next speaker is Chief Medical Officer for the Intel Corporation. You might of heard of them. Umm and I was I was surprised that Intel had a Chief Medical Officer and I am personally really excited to hear what uh Dr. John is going to share with us. So please join me in welconing- welcoming to the st- uh to the DEFCON stage Dr. John Sotos. [applause] [indistinguishable shouting off mic]
Thanks I will [voice off mic] Thank you very much. Did somebody say break? 
>>[from crowd] BREAK!
>>I’m a cardiologist I'm not gonna destroy brain cells. [crowd laughs] It's against my professional ethics. [applause] So uh it's an honor to be here today um I hope you get something out of the talk at least something to think about but first I of course I have to start with disclaimers. What I’m gonna talk is not necessarily the official position of Intel or the Department of Defense. I’ve got a Department of Defense affiliation too which I’ll go over. So anyway with that let's get going. I’m gonna use my um pointer over on this side I hope everybody uhhhhh ok. Well. Here we go. 
So first uhh let's go back to 1978 when I was a college student uhh in about  this part of the country. [clears throat] And umm I got sick really sick I was sick as crap. I was laying in bed shivering [clears throat] excuse me, my teeth were actually chattering and every muscle in my body hurt. Wasn't just me it was everybody it seemed like. Well almost the professors didn't get sick. Just the students. And it wasn't just my college. At the Air Force Academy 77 percent of the recruits got sick, the cadets, sorry. It wasn't till 30 years later I found out what had happened. In 1918 that's when the story starts, a virus got from a bird into humans and that killed about 1 percent of the entire world's population in 1918. And that virus and its descendants stayed circulating in the population for another 40 years or so. And you know how there's an influenza every winter and there's a different shot you get every winter the virus changes little by little so over the next 40 years that virus hung around. And then in 1957 it went away it was replaced by another influenza virus, an H2 virus, and then in 1968 another virus replaced that. So far all is normal.
In 1977 though, H1 came back. And in fact it just wasn't a random H1 it was almost exactly identical to the H1 from 1948. And that's why the professors didn't get sick. The professors had all gotten sick in 1948 cause they were older. It was only the students who had never seen this virus that got sick in 1977. So how did that happen? Because these viruses change you can't get a 30 year span in time when a vi- when an influenza virus doesn’t change. And so the common consensus is the reemergence on this virus in 1977 is unexplained and probably represents reinduct- reintroduction to humans from a laboratory source. 
What does that mean? It means something got out of a lab. It got out of a lab and it came halfway across the world and it got me. This is where this virus first appeared in the wild, somewhere in northeast Asia. We don't know if it was the Soviet Union or China. But it went worldwide. And it didn't affect the old and the infirm it affected the healthiest people; the college students. So you’re affected by this too because ever since then, the descendants of this H1 virus have stuck around and this H3 virus has stuck around too. So every year when you get your flu shot and I hope you do, you’re getting immunized against something like this and something like this. So that's the one of the motivations for this talk. Stuff that happens in labs can have worldwide reach. And we know that micro engineer- that engineering of microorganisms is happening every day today in laboratories. The question is: What is the potential when malicious engineering of organisms starts? And how do we defend against it? And I would- claim that only massive preemptive development of counterhacking bio technologies can save the world. Cause I think the threat is that grave. 
So who am I? As I mentioned, I’m a cardiologist who doesn't drink much. I work for Intel uh I’ve been in the Air National Guard for 30 years uh mostly as a rescue flight surgeon. I’ve been programming since 1970 uh I don't get to do much of it at Intel unfortunately that was my first computer that I ever programmed there. Am- am I a hacker? Do I deserve to be here? Well, I wouldn't presume on the computer side but I have an interest in what you might call diagnostic hacking. Hacking the diagnostic process. And I wrote a book on that and I was a professor for a long time but uhhh you probably know me if at all from my work on House for 6 years where I was a consultant. And also on uh Torchwood Miracle Day, and umm I’ll be consulting for a new show this fall on ABC called the Good Doctor. So tune in Monday nights. [crowd laughs]
So today I wanna talk about uh epidemics and bioweapons. Talk about digital biology. Uhh show how exploits where they’ll come from and then uh conclude with a few uh reflections. 
So uh bioweapons are actually amazingly effective and they date from ancient times. The Hippocratic Oath that all us physicians take that started because there was a doctor who told his his military commander how to poison a village and uh over run it. So uh in the middle ages the black plague killed about 25 percent of Europe's population. Smallpox killed about 95 percent of the Aztecs, that's why nobody in this room knows any Aztecs. Influenza as I mentioned killed uh 1 percent of the world's population and it went everywhere there were only 2 places on earth it didn't go. And even today we have an epidemic of small headed babies. I mean who would have thought such a thing is possible it turns out the Zika virus can do that. And uh no less than Bill Gates says “Of all the things that could kill more than 10 million people around the world, the most likely is an epidemic.” So uh we haven't even started to see unnatural epidemics, just the natural epidemics. So this I saw an exhibit in uh Oxford England this is from a book published in 1625. And um when the black plague came through and so here you have obviously death and the population of the town we fly, they try to get out of the town you can see soldiers stop them. But even though they try to flee, death follows them and then we die. Here are the coffins. So that’s what we’re up against. 
And the effect of bioweapons can be long lasting. Um the Brits tested some anthrax in the world war 2 and the island where they tested it was uninhabitable for 50 years. Malaria has affected the evolution of multiple human genes, I’v got a gene that that I have only because um my ancestors survived malaria and they were good at it. And about 8 percent of your genome started out in viruses. So the problem there the reason you haven't heard much about bio weapons when we talk about war fare is they've been held back by a pretty severe limitation which is a potential for blowback. If I use a bio weapon on some adversary across a border, that epidemic is probably gonna spread back and get me to so that doesn't make it a great weapon. There are international treaties that outlaw bioweapons but you know treaties are easily broken. So lets get now into um the exploits and and how they might work and how to think about them. 
The first thing to realize is what's called the central dogma of biology. You start with DNA and you make RNA out of that and then you make a protein. And the proteins are the real work horses of of life they do various jobs. Now if you wanna make a medicine today, you’re usually making a medicine that attacks the protein and for you know just convenience we can call that analog therapy. Uh you know you des- design a chemical that fits into the protein or does something to it and this is pretty difficult and its imprecise, there’s lots of cross reactivities across different proteins and that makes it hard to design safe and effective new medications. Its why drug companies spend so much money developing drugs because it's so difficult. 
So tomorrow though I think we’re gonna see digital medicines, and we’re starting to see a few. Remember RNA and DNA are digital programs, they’re written not in binary but in uh quaternary code, A C T G A. Umm and there amenable to digital manipulation which means you can reprogram them and so this is going to allow an algorithmic or a digital design of medications. And the Cancer Moonshot which uh vice president Biden uh of course pushed is going to really uh drive these new technologies to manipulate DNA. Because cancer really is a disease of DNA. And if we get a good cancer uh mechanism er uh uh mechanism to combat cancer going uh using this root it's going to be exquisitely persp- sp- specific. 
Because cancer cells are not that different from normal cells. OK and just some of the th- um catch words you might have heard, some of these digital DNA technologies, there’s something called RNA Interference which won a Nobel prize in 2006. You’ve probably heard about Crispr cas9 and uh for sure there will be a nobel prize for that some day. There’re things called gene drives. Um spreading these um DNA programs can be done through measles virus let's say it's unbelievably contagious. Um and even nanodiamonds can be used to get uh viruses, er uh DNA into you. So what you could do with this sort of t-uh digital approach to DNA and RNA is you can do things like program in an if/then statement. And this is already happened uh in fact somebody programmed a 5 part predicate in here for the if statement. So far nobody I don't think is working too hard on th- or is is succeeded in the deploy payload part but the if part is is very uh well along the road. And this sort of construct is gonna be the key to biohacking.
 So um I’m from Intel we talk a lot about Moore’s Law, but uh biotechnology is blowing Moore’s law out of the water. Uh this graph here shows how much it costs to sequence a human genome, going back to I guess that about 2000 I cant read it. Uh and this uh lines this straight line is Moore’s law and this is the cost of genome sequencing and you can see it is dropping way faster than Moore's law. So in 10 years we might be talking about a 10 dollar genome sequence. And uh you know these sort of uh  if/then statements are gonna continue to um get better and better and so this whole talk was prompted by the question: “With this kind of exponential increase in biotechnology, with new things like Crispr, where is bio technology going to be in 5 years or 15 years?” And it’s kinda scary because defensive technology always lags offensive technology.
So um you know the Cancer Moonshot I would propose is dual use. Just like nuclear weapons and nuclear power are two sides of the same coin, the ideal cancer treatment some day is gonna be: the doctors gonna biopsy your tumor, get a sample, send it down to the lab, the lab will figure out the genetic syndrome, or the signature of your exact cancer tumor, then somebody will build a virus that using that if/then statement only targets the cancer cells in you. They’ll put that virus inside you, you’ll feel like you have a cold for a few days, and then that virus will go to work and tha- that’s because the virus is going to work destroying your cancer, and then you’ll wake up cancer free. That’s a pretty good deal, we all want that to happen. But notice this exquisite targeting overcomes that big drawback against bio weapons. 
So the new technology is gonna allow incredibly targeted bio weapons. So think about 3 different axis. Who might you target? A specific tissue? That’s what cancer does if that that cancer therapy. And if you wanted to edit an embryo uh as is just starting now, uh that will certainly come along too. But you could also target a family like the royal family, you could target a whole group of people, you could target an entire species so that's the who. Then there’s the what you know if you’re doing cures that's great but you know you might be making life inconvenient for somebody you might be giving them a disease, or you might be killing them. And then there’s the when. And uh there’s a very important factor here I’ll talk about later too. So if you were gonna cause death in an individual well we’d call that assassination, or murder. If you were gonna cause death in an identifiable group of people, we’d call that genocide. And you might say “Well, why would we ever want to kill off an entire species?” If that species is the malaria parasite, or if it’s the mosquito that carries Zika virus, or Yellow fever. You'd want to kill those off too. So certainly this part of the graph is uh in the sights of scientists working today. And so this talk is about what happens in this part of the graph. The technology is inevitable, let's remember that, everybody wants that cure for cancer. And so they're gonna be you know I think thousands of people down in the basement of hospitals doing these viral manipulations. This genetic program and that's a lot of power to put into those people. So what could happen?
Umm let's talk about that. So first um let's talk about that “what” axis. The damage people could do. So I’m gonna use genetic diseases as a guidebook for exploits. And here I define disease as any kind of abnormal function in your body. So an exploit is anything that includes a genetic that induces a genetic disease. So here’s an example. Here a rare genetic disease called Xeroderma Pigmentosum. Add this uh little girl has it and it's a defect of DNA repair that arises from a variant in the XPA gene, one of the 25 thousand genes that you have in your body. And these people are intolerant of sunlight. Every time sunlight hits your skin, it damages DNA. But your body repairs it. These people don't have that repair capability. So when they go out in the sun they will blister after a few minutes. OK now if you’re hackers and you could deploy something like this worldwide, you know, why might you want to do it? Think about that for a few minutes and and I’ll come back and and give you a suggestion.
So there’s lots of potential in our hu- human genomes to cause this kind of problem. This is um my mentor at at Johns Hopkins he was a terrific guy and back in the 1960s he started collecting every inherited trait that the profession of medicine could identify in humans. And uuh this is my copy of his book. Uh in 1990 that's the 1990 version and if you’ve read Moby Dick you know you can see that it's way bigger than Moby Dick. And that was in 1990. And since then the book got so big they couldn't publish it in hard copy anymore and this only goes up to 2004. And uh you know I think they just gave up counting at that point. So we know a lot about what happens in people's bodies. So within this book this big book of uh genetic umm ummmm susceptibilities in human beings there’s a lot of stuff. And in fact there’s some stuff worse than dying. And I and I call that hell. And and you don't want to know what that is. What could happen. So the question is really: “What happens in these 2 places?” So let's give some examples. So let's say you were a passionate animal rights person. And you didn't want people eating animals, you might be uh a passionate vegetarian as well. Well there’s a disease called Ornitheintranscarbamolise defficiency which is in that big book, and it arises from a variant in the OTC gene. So if you could spread if you could mess up everybody's OTC gene across the world, all the sudden, almost nobody in the world could eat meat. That’s something that as a an extreme vegetarian you might uh want to have happen. Let's say you believed in sexual chastity and you really wanted to punish people who were a little profligate, you could make everybody in the world hyper susceptible to Gonorrhea by messing with the C5 or the C6 7 8 or 9 genes. It’d cause lots of other problems, but you could you could make people really susceptible to Gonorrhea. Let's say I didn't want anybody at DEFCON to do a shot. And I wanted to make them intolerant to alcohol. I could screw with their ALDH2 gene or to go back to xeroderma pigmentosum if I wanted all women to be veiled when they went outside I could design something that would mess up the XPA gene only in women. Or if I wanted uh to uh blur the some of the distinctions between races I could um distribute genes or uh vectors that would cause skin color to change. If I’m a pharmaceutical company uh and I have a drug that treats some genetic illness, how great would it be if everybody in the world had that genetic illness? That would do wonders for sales. And let's say um you were once turned down for the astronaut program because you were colorblind, I don't know who that might be. But you’re looking at him. [crowd laughs] And um [more laughter] let's say you wanted everybody in the world to be colorblind, then they couldn't turn you down. [crowd laughs] So there's more [clears throat.] Let’s say I just don't want to um hurt people but I want to target national economies I have a a nat- uh an enemy an adversary and I know that it would bankrupt their economy to take care of epidemic expensive diseases. So there are ways that I could give people cancer, Parkinson’s disease,  Alzheimer’s disease, cystic fibrosis, I could give everybody pain all the time. Or I could make them insensitive to pain, which, trust me, uh causes a lot of problems. I could immuno- compromise people or I could accelerate aging. Umm [clears throat] let’s say I just wanted to impair uh do things to impair workers. I could give everybody narcolepsy. Um the face blindness syndrome is probably genetically deter- uhh mediated no one's discovered the gene yet. Or if I wanted to give a whole generation of children learning disabilities so um some country couldn’t compete against me, I could do that. [breathes][sucks teeth] Let's say I just objected to se- to a life style. Um you know I could make everybody deaf or blind and look at this there are 540 genes that goes into our that go into our hearing process and 600 genes that go into vision. Could make everybody night blind, interfere with taste and smell, and destroy uh a big industry right from that. There are genetically des- er mediated diseases that make you die from excitement, these are cardiology diseases actually. And then there are diseases of physical fragility. OK, we’ve got a few more. Uh epidemic micropenis, ummm [crowd laughs]. Epidemic erectile disfunction, or you could make everybody uh hyper libidinous. You could I think I couldn't quite find the gene but I’m sure this is genetically determined somehow, uhh you could fix it so a nation just became a nation of sons or a nation of daughters. Um there are reports that uh certain organizations uh were looking to try to change the sexual preference of their adversaries and uh you could sterilize an entire population. 
Uh let’s say I just wanted to mess with some politicians and you know I could make somebody go totally bald or I could give them a really bad fishy odor and that is actually an incredibly tough disease to have. It's caused by this TMAU gene because there's no treatment for it and it's impossible to disguise the odor and those people have a very difficult time socially. Uh I could give somebody intractable diarrhea, cause them massive weight gain, or um you know in some forms of Tourette's syndrome which is mediated by this gene, people uh involuntarily uh emit obscenities, so that would be uh interesting in a politician. I put this in before uh Mr. Scaramucci started [crowd laughs] his work, so. [laughter and applause] We might have to cross it off the list. [crowd laughs] OK that's the what. 
Umm we talked a little bit about the population um I would recommend keeping your genome secret, ummm so you know we talked about the royal family and you know you could get just uh DNA from one member of the royal family and and you could then target the whole family. Um you know this word “race” is uh very squishy, but if you use it to mean any observable physical characteristic defines a race, then um you know it it becomes possible to think about in a sort of rigorous way. So for instance um there’s a gene called EDAR and 87 percent of asians have that. And um you can exclude asian descent in nearly 100 percent of europeans and africans by looking at this one gene. So if you wanted to do uh racial elimination um EDAR is a good place to start if this is your target. So obviously this is very dangerous if used by hate groups. And then there are easy ways to identify uh genetic males and genetic females, which of course is genetic. Umm you know identifying ethnic groups is a little tougher. Uh you know Tay Sachs disease tends to run in umm the Ashkenazi jewish group. But even there it its not very good uh quote unquote, if uh you’re using it to target them. So that may be sort of a a blessing you know interbreeding uh is is helpful. Umm and we talked about species, uh you know if you're a dog person and you want to wipe out cats you can do that or vica versa. Uh you could wipe out uh food sources. And you know if you um wanted to kill a wine crop uh if you’re from Napa and you wanted to kill the French wine industry uh you might be able to do that. 
So let's talk about time. Um you know I think it’s possible to build binary weapons, so you insert one thing but the bad thing doesn't happen until a second agent is exposed. But the really interesting thing about time is yeah I could do something to everybody who is living here and make some of you sick or whatever but if we can get that gene change into the cells that make sperm and ova, then it's in human beings umm in perpetuity until somehow they all get fixed. Umm so this is just the RNA and the DNA part you know there's a whole nother field called epigenetics and epigenetics is way the environment umm signals in the environment are transmitted to your genome. So for example if you moved to Denver where the air is thin where there’s not much oxygen, somehow that signal has to get from the oxygen in the air into your genome so that you can make more red blood cells. So there’re gonna be I think a whole host of epigenetics uh exploits that uh I really haven't started to think about too much. Umm and then uh targeting based on the microbiome uh would also be uh possible. So uh to summarize, you know a lot of this is possible now, some of the these engineering um angles but it's very difficult. And so not many people can do it. But um with the progress that we all hope the cancer moonshot will make, its gonna get easier and easier to do these kind of things. And then when it's to the point that thousands of people can do this sort of stuff, we have a real problem. So um my first uh statement to you is: don’t do this stuff. Umm there’s there’s  just no point in doing that. Yeah we know it can be done, you’re not showing anybody you’re smart if you do this. Uh but if you have a chance to talk about this problem with policy makers uh do that. Um you know if this group gets scared because not much scares this group, but if this group gets scared, I think that’ll get their attention. Or get involved more directly. Go into bioscience yourself and uh try and put your digital way of thinking to work in trying to build the defensive technology for this. I don't have all the answers that’s for sure, so we need a lot of help. 
What I would tell the policy makers to do is to, and and this is very difficult for them its essentially saying “You have to start working on defensive technology for an offensive technology that isn't here yet.” But the goals are really quite laudatory in a sense that the first step is to figure out how to cure every infectious disease known to science. And you’d want to do that using some sort of digital technology. And that alone would create the largest amount of wealth that I think uh the world will have ever seen when you think about how much infectious disease costs mankind and and how much people would pay uh not to get infected. Um but that's only the first step um because there're gonna be unnatural infectious diseases that people will build and so we’re gonna have to develop an infrastructure that will be able to detect new agents and then characterize the infection, devise a countermeasure, produce the countermeasure, produce the countermeasure in millions or billions of doses and then distribute it across the world, and oh yeah, probably within a month or a couple weeks, of the time the disease is detected because that's how long it takes you know a disease that we get here could go everywhere in a couple weeks. And you know, not too long ago uhh in- infections were of a magnitude that we can hardly believe today. In 1947 New York vaccinated 6 million people in 4 weeks against smallpox. So I don’t think this is necessarily impossible, I think it's harder than a moonshot, it’s like a Mars shot though. 
If you’re interested in this and you want to get more motivated or maybe I should say if you’re not interested and you think “Uh is this guy up here talking crazy?” um this is a novel that was published in the 1990s and um it is a terrific novel just on its own merits but eh in the introduction the author says uh “All the science is in this novel is true except for 3 things, and I’m not gonna tell you what they are.”  So that's what the author said, so uh read it and and have a good time and think about it. And then finally thanks for sticking around today toward the end of the conference and hearing um remember that uh as we leave the conference we’re gonna go back all over the world so it's a pretty good thing that we're not carrying any bad bugs with us. Thank you very much. [crowd applause]
Uhh, John, do I have time for questions? Where's John?
>>Yeah you do.
>>OK. Do they go to the mics? Whats they uh..:
>>Yeah
>>Routine here?
>>If you’ve got a question, come to me. 
>>Thanks. So I think one of your colleagues just tweeted that you said “Don’t share your genome.” And I’m wondering how you see reconciling that with doing the research that's needed to better understand genomic medicine and also to treat people genomically, presumably for me to get that precise treatment I’m gonna have to share my genomic information. 
>>Yeah, if you’ve got a genetic disorder umm you have approximate threat to your existence and to your life and so there the balance would tip towards sharing the genome so thank you for asking that. That’s worth a clarification. Everything in life is a risk benefit balance and so I would say umm don't share your genome without reason. Without good reason. 
>>Hey um so attribution is a famously challenging problem in computer security, determining the wh- the origins of malicious software. I was wondering if there is any current work that you know of going on around um watermarking these modifications to genomes or other techniques that could be used to trace their source and their progress. 
>>I I couldn’t understand that question. I mean
>>Oh
>>>Talk a little slower. 
>>Sorry, heh [crowd laughs], um so so attribution is this famous problem in computer security like um when a…
>>Oh
>>..malware spreads around the world...
>>Yeah
>>...where did it come from? 
>>>Yeah
>>...was it Russia? Was it China? You know?
>>Yeah
>>Um and so what I was wondering is is there any current work happening in genomics that you know of umm around developing ways to uh watermark or otherwise umm annotate these modifications so that you can determine where they came from and how?
>>Uh that's an interesting question . So uh no I haven't heard of that you know um the whole science of epidemiology uh focuses on where did things where did infections come from? How did they spread? And that's been very difficult over the hundreds of years that epidemiology has been a science. And even when we look in genomes we can get some help so uh but but not the kind not the kind of help that you want. So I can if I gave you influenza um the match between our two viruses would be like 100 percent minus epsilon. Some really small number. But if he got influenza from somebody in Portugal it would still be Influenza A an H1 or whatever but his difference between your 2 influences would be bigger. So you can do some sort you can do tracing that way uh but you can’t go and say “Oh it’s started in you know the Amazon jungle.” or anything like that.
>>[off mic] If you could [unclear] you could like actually code that genome. Like you could code [unclear]
>>So the question is about in coding uh signing modifications uh that you put into the genome sure. There’s lots of room in the cell you can add a few more uh DNA base pairs um that you know somehow you would claim or something like that and and put a signature in there. 
>>Thanks.
>>So, I’m thinking a little bit about data archiving. In long term data archiving large amounts of data um one of our primary concerns is bit rot. And and so we so we check that frequently and then readjust it from available sources. I’m wondering if it- it would be reasonable to think about the DNA umm in an archiving maybe personal archiving sort of like backing it up where then with frequent checking you could find something that had gone awry and then essentially restore it by doing it. 
>>Yeah. [chuckles] Do you have an identical twin?
>>No
>>Too bad. [audience laughs] Umm so [more laughter] Uhh yeah identical twins are are great great for that kind of backup well. You know uh for cancer umm eeh you know there there’s another way which is um se- er cord blood. You know when babies are born they have the umbilical cord and nowadays you can snip that umbilical cord and put it in a freezer and uh that thing is that cord the blood in that cord is loaded with stem cells. And ummm you know if you get a leukemia later you can go back to the cord blood and you have a sample of blood there that is essentially the backup you’re talking about there’s no leukemic cells in it. Umm you know that was sort of the rage a few years ago I don't even know the technology may have passed that by, I can’t say, but you know just from a um from a sort of IT perspective if you put your entire genome in some digital format and put it on a hard drive or something umm the question is how do you restore from that backup? So you know they're several trillion cells in you and that's the worst case scenario is you’d need to get that backup into every one of those cells but if it was the sort of thing where you got leukemia and they were saying: Hey we can’t find any non leukemic cells in your blood to do a transplant, a bone marrow transplant on you maybe someday in the future you could go back to that digital backup ,build some cells with your correct DNA and then infuse those into you. 
>>Regarding your uhh device to work on contra measures umm working on contra measures means to go through the same technology because if we need to fix the- um actually is easier to break then to fix so the whole genetic therapy it- it- it's a bit old and but it's really declined because its really very difficult to fix stuff...to break is...
>>Yeah yeah that’s why defensive technology tends to lag behind offensive technology. But I would say there there’s still ways you could do it. So let’s say Crispr turns out to be the former Director of National Intelligence for the United States called Crispr a weapon of mass destruction. So if just working on anti-Crispr technology for instance might be a way to go. So that you know whatever program is embedded in some sort of Crispr implementation if you ca- had a way to kill the Crispr part umm that would be potentially in some situations useful. Umm and the same thi- you know but I think Crispr might be kind of easy. Other things like RNA interference that might be really hard to develop countermeasures for that. But that that’s what I was thinking when I made that statement. 
>>Hey. So I’ve noticed a recent rise in popularity of these ser…
>>Talk slower, ‘cause
>>Oh my apologies
>>Yeah, ‘cause…
>>So I’ve noticed a recent rise in popularity of these services where say they’ll send you a vile, uh you spit in it, you send it back, they give you information on your genome. I’ve seen uhh a lot of uh educational channels on youtube encouraging people to go out and do this. Uh and it seems quite an interesting thing to do if not something uh information that would be good to know. However I worry about the security of these sort of uh, of these sort of organizations. Are they selling my data? Um what government organizations? Uhh wh- what would your advice be for the security of these? Where do you think uh how do you think we could make sure that uhh our data isn't being sent out? Uh should we discourage family members and friends from using these? That sort of thing? 
>>Yeah I think it comes back to a bit of what the first questioner asked was the risk benefit trade off. So if you have a medical condition where you think your care could be improved if the full genome were known, ok I can see that. But if you’re trying to figure out whether you’re Lithuanian or Greek or something else, you know, who cares? Ummm [audience laughter] Soo [scattered applause] [laughs] [applause] And you know I had a friend do that and he found out that his father wasn’t his father so you know you really can't get great news from it. But you know [audience laughs] ehh eehh it what I worry about as a physician is let's say there are 5 thousand diseases that are mediated by genetics. There are probably more but lets say there are 5 thousand. And let's say some day you get a report back that lists your risk for all those diseases. Well you’re gonna be at above average risk for 25 hundred diseases, just statistically. And so if you come in to see me and I’m your doctor and you start talking about 25 hundred diseases that you’re at risk for, I have no time to talk to you about anything else. You know, I know that getting your blood pressure down by 10 points is gonna ss- extend your life by so many years or helping you to quit cigarette smoking is gonna extend your life by 10 years. So you crowd all of that out for this information that really isn't helpful. So uh that’s that’s what I would think about. Medicine is very aggressive in the sense that you know if if we have a way to decrease your risk of disease um we’ll develop screening technologies around that. But if we don't have any way to decrease we don't have a way to decrease your risk of rheumatoid arthritis your risk of psoriasis or anyth- anything like that so there's no point umm in a preventative way of of coming of getting tested for that and coming to talk to me about it. 
>>Thank you
>>Mmm-hmm
>>So the topic has come up a couple times now that…
>>Slower, yeah cause my there yeah
>>That it’s easier to crash a human than fix it. {clears throat] The number one problem with gene therapy that we’ve observed from our historical data is random insertions into the genome that lead to different errors that are actually much more catastrophic than the error you were trying to fix. So is anyone working on how you would be able to have a defensive strategy to not have that happen? Because in its current state gene therapy is kind of like playing russian roulette. 
>>So [clears throat] um let’s go back to that big thick book that I showed a picture of. Most of the diseases in there are single gene diseases. Uhh like that xeroderma pigmentosum that's one gene. But if you look at what humans tend to suffer from: obesity, alzheimer’s disease, anthroscoladic heart disease, you know the common stuff, those are multiple gene diseases. And so umm it's there are almost like two different sets of targets. The single gene diseases we understand how to find em and we understand how repair would occur theoretically. The multiple gene diseases and I think this is what you're saying are much more mysterious .You know if there are 640 genes that go toward making your visual system work correctly, uh thats a bad example. If there are 50 genes that are involved in umm atherosclerosis you know when you start tweaking one its gonna have a bunch of other effects and your not gonna maybe find a big effect from any one gene so now you're talking about diddling with multiple genes and so that's really hard and that's gonna be quite a ways in the future. Is is that what you’re asking about?
>>Uhhh 
>>Cause this you you mentioned like stop codons and things like that that's sort of the single gene disease. 
>>Yeah it doesn't matter how many genes are involved in the disease. If you introduce something into the code that was external so you downloaded information and it went to the wrong file... 
>>Mmm 
>>...and it caused an error that was significantly worse than the original error, the best case was a single gene gene therapy approach the genes randomly inserted into the genome and caused leukemia in most of the patients in that population. 
>>Yeah so that’s gonna be part of the safety and efficacy evaluation of any new therapeutic that comes out. So you know the first gene therapy efforts umm didn't work. You know it it caused harm but lessons were learned from that and so it’s getting better. So uhh like an earlier questioner said, it’s always easier to break something than to fix something. And any attempt to fi- every attempt to fix something you know you have to be prepared for unintended consequences. And so its just gona be a matter of ref- refining techniques. Your point is well taken uhh but medicine is is pretty used accustomed to trying to work out what the adverse effects of any therapy are.
>>Mmmm one final comment on that. Do you think it would be a bit more responsible to target something that's not the equivalent to the source code? 
>>Umm well that might be like the microbiome because if you screw up your microbiome in some terrible way you can always flush it out and load and reboot essentially by eating the feces of some other person and uhh you’re good to go. [crowd laughs]
>>That would be a much better approach. 
>>OK I’ll put you down for the feces eating. [audience laughs]
>>Huh that was the most scientific way I’ve ever seen somebody put down a question. That was amazing. [audience laughter] 
>>Uh close- eh closely related to the uh last question, eh would there be a way or is there any research going towards ehh basically locking the editing of the genetic sequence. 
>>Yeah
>>Turn evolution into something purely directed that would be the int- the there would be no more natural vectors of evolution if we did that but it would secure the uh the the the whole code. 
>>Uhh I that that’s a very interesting idea. I I don’t think there’s any work going on that. Uhhh I wouldn’t know how to implement that. Umm you know there there’s 6 uh billion uh between the 2 parents contributions that’re fff- billions of base pairs and so you know we have uh the 2 rungs of the double helix which uh provide some measure of error correction in there and it is possible to build a triple helixes using not DNA but there’s now uh new kinds of NAs, XNA, and you can build triple helixes so you could actually umm increase the error correcting capabilities but then I think you’ve got something that’s not a human at that point. Because ummm it's just so different from us. 
>>[inaudible] ...we gotta get through...ok
>>Hi there how are you doing? Uh so I have 2 questions, 1 is pretty quick. Umm you talked about changing some physical traits about people to target certain groups. But how complex do you think we could uh develop that technology? Such as if someone has a birth defect where there fingers may not grow properly do you think we would be able to correct that and then through there uhh I guess preteen years when their bodies are developing more they would be able to grow proper fingers, I guess? 
>>So uh there are a lot of people working on that. Umm you know there are some malformations that occur not because of problems in the genetic code but because of environmental influences. So if you have an encounter with a chainsaw you know you could end up with a uh an abnormality in your anatomical structure. So um you know I had a friend who um found out that the tips of your fingers, the skin on the tips actually regenerates and so he took this as part of his research to figure out if he could expand that regenerative capac- uh capacity beyond your fingertips to the to other parts of your body. Uh he didn't win a Nobel prize so I’ll I guess it didn't quite work as well as as we might hoped. But there there’s a tremendous amount of work in that going on. 
>>Thank you
>>We got time for one more...
>>Okay
>>...question. Sorry. 
>>Hey
>>Ex- excellent talk, I just wanted to start by thanking you for scaring the ever loving s**t out of all of us.[applause] Umm because frankly that needs to happen if we’re going to avoid this terrible future that you’ve described. Crispr is the new kid on the block but I think of Crisper as like a root kit that's just really good at doing advanced things but it's the spreading mechanism that frankly mother nature has been giving us for millennia that's the real problem to address. So my question is umm are Intel or the DoD or other organizations that have the resources to tackle this developing the equivalent of biological intrusion detection systems that can sample airborne or or human spreading epidemics through urban population centers to very quickly identify that spreading vehicle?
>>So that’s a great question. Umm there there’s sort of 2 mechanisms one is the science of epidemiology which I’ve already talked about and of course that is limited uh you know it started in the ss- 1700s or 1800s and we still use the basic same mechanisms. Um on the technology side the Centers for Disease Control has something called a biowatch program which is these today these big carts umm so the San Francisco uh San Francisco area hosted the super bowl last year and on the sidewalks in San Francisco these carts appeared and they have you know sort of a smoke stack and they basically sniff the air and they bring the air into the business part of the cart and with today’s technology it's just deposited there and a human comes by, picks up the samples, takes ‘em back to a lab and does some PCR some genetic analysis on it. So the CDC tried to develop a ss-second generation of this cart and the program was not successful. And so right now it's in hold as to what to do. Umm you know my idea is that umm when you go into an emergency room and they draw some blood from you and send it for the lab, the blood in that tube uh 90 percent of it is thrown away cause it's more than they need. So why not take that blood and throw it into a big vat and once a day just sequence the crap out of all that blood that comes in and send that immense amount of data down to the CDC in Atlanta and that would give them really I think a national picture in real time of what’s coming into the emergency room at a genetic level. So I I think that would be kind of a first step on um a new generation of bio surveillance. 
>>Uh one one for you this is quick. Is it lupus?
>>Ha ha ha ha. It’s never lupus. [applause]
>>Yes!
 [audience laughs] [applause]