Dr. Gordon Sanghera, Chief Executive Officer, Oxford Nanopore Technologies  
Interviewer: Dr. David B. Agus, Ellison Institute for Transformative Medicine


Brainstorm Health 2022 - The Long Read: Gene Sequencing And The Cloud 

Conferences

Brainstorm Health 2022 - Gene Sequencing And The Cloud 

alright. Gordon? You ready? Born reddish. Right so when I go to my doctor's office now and I go in they stick a needle in me, they draw blood, they check my blood pressure, they look at me and then they send me home and a couple of days later they call me with the results. And then if they did my gene sequencing it'd probably be 34 weeks later they called me with the result of a gene sequence. You have pioneered a company that is kind of gearing to change this. So what is Oxford nana poor. So um our mission in healthcare is to deliver a lab of the future. And as it happens there's one in my pocket here not my shirt pocket famous. Um So this is a DNA sequencer sequencing as you all know because of the pandemic is the source code of all living things. And it can decode DNA. RNA in any living thing now what it can do it can do it in real time and it's very low cost. So it's affordable and it allows the accessibility of data generation near the patient in real time. And that will be game changing for patients. Same day results you can imagine I'm seeing a patient with breast cancer and she has a triple negative breast cancer the most aggressive format. And I want to know if her B. R. C. A. Or breast cancer genes are mutated or not because that's gonna dictate whether she has a mastectomy that's gonna dictate which cancer drugs I use and how she's treated and you can imagine she's sitting there on the edge of her chair waiting to know in her care and I say listen I'm gonna draw a tube of blood. Three weeks later I'll give you a call and then we'll figure out your treatment. So you're telling me that while the patient is next to me I can sequence that gene and get the results back and be able to tell her what her treatment is going to be. That's right Using even using today's technology which isn't tuned we can do it in a couple of hours But this has the potential to do it in that 45 minute turnaround time that you need in the consultation room. Alright can we see it? We can't indeed done so to really bring it to life because it all sounds like something that's left 100 light years into the future stan is going to start a sequencing run of a cancer panel um a lung cancer panel and we will see the results during the course of the experiment and we'll talk through that a couple of times. So over to you dan what are you gonna do? What we've got is a sample of human DNA. Here and we've amplified 16 small regions of the genome which we're going to sequence today. Um I'm about to so these are a panel of genes including things like E. G. Fr and rest. I'm gonna load this sample into the sequencer in the next few minutes you're going to see what it looks like to do DNA sequencing with real time data analysis. So while dan is doing this little bit of manipulation which for now requires a 16 year old train technician to use a pipe at we have a vision where that will all just be drop of blood or urine or whatever the biological sample is like a like a bit like a lateral flow test. A very sophisticated version of that that will allow gene reading and instead of just providing a very simple absence presence it will give you a very sophisticated output. And as I've said it's about the data for the patient. So this isn't science fiction. I mean you're actually using this and scientists are using it across the globe now. What percent of covid sequences In the world on this platform now. It was about 15% over one million genomes. Covid genome sequence with this. But what's significant is it was done in countries lower middle income countries where they cannot afford to build multi million dollar centralized infrastructures to do sequencing. And that's the real game changing element. It really opens up and democratizes access to DNA RNA information. Where does the data go? So you're gonna show us how it works and we're going to learn that in a minute. But when the data is created where does it go? So it's the data is streamed in real time and we'll show you a short short video in a minute that goes up into the cloud. So one of the things we're excited about working with your institute and with oracle is to use hippo compliant pipelines so that that can in real time stream. And we're off and that data in real time, we can see what's happening to that patient. We can get confidence schools on the risk levels of a given cancer that you're trying to look at based on the test you're running. All right. So, we're seeing a bunch of channels and a bunch of dots. Right? So what that is over there. So inside here there is a non apple, The Nano poor very small hole, 500 of them. And what you're seeing there are 512 individual nana reactors all going to be sequencing in a moment. The system is checking itself. The sample down added is now slowly reaching the surface of the platform and you can see the green emerging. So that is now strands of DNA actually going through these very small holes. And um, to sort of, that sounds all very difficult to comprehend. I've got a video. I'd like to show if we can play the video and I'll talk you through that, I'm actually gonna go over there and give you a life run through. But it really explains what we have. So inside we have a flow cell, this is where all these 500 channels are. So the channel this is the DNA prepped ready for sequencing. The channel which you're going to see in the moment is a very a protein with a hole through it. It looks like a mushroom. And there are hundreds of these channels which is what you're looking at here. It looks like a protein like a like a mushroom with a hole through the middle of it. And as it happens it is perfectly sized to take the iconic double stranded DNA that you're seeing here and to have a motor that Singulair hates it and it passes it down the whole and as it's passing it down the hole the G. T. A. C. Give you slightly different changes in current. Now this does sound like something out of the future. But what that means is you get a signal out. We use machine learning ai to convert that signal the sequence and it continuously reads strands and we'll go back and look at some of those during the course of the experiment. And that basically is nana for sequencing. It allows us in real time for the first time ever to generate this data And then you can do one patient then another then another. So that random access makes it look and feel like a traditional diagnostic workflow. So right now you need a PhD you need a $75,000 machine and you need a lot of bio statistic power to understand and this is transformative and what it can do. So tell me about things like you know public health. If I'm a country, if I'm Indonesia and I've got a big tuberculosis problem now, how can I use a technology like this to help? So the the paradigm of care for tv today is point of care testing which gives you absence presence. The problem is in countries like Indonesia India there are high levels of drug resistance, multi drug resistance and what we've been working with with the World Health Organization and find a non for profit diagnostics um outfit is to show that we can for very low price points, tens of dollars, we can find all the drug resistant profiles. And that is a differentiator between using next generation sequencing, which people talk a lot about versus a diagnostic without prior knowledge. You can find all those drug resistant strains because you sequence the whole genome and that is game changing because what happens now is you've got a drug resistant strain, you're given the mad, it doesn't work. Iraq's spreading and that just is never ending upward spiral of infection. You know, I don't remember what exactly 68 weeks ago we were gonna have a call and you told me, hey look at new England Journal medicine. You know this week, I think you'll be impressed. And it was about a group at stanford um that had Children who were born uh with uh errors. Um and they were had issues. And literally you sequence these Children overnight. Um so tell us about that story. Sure. So what you're looking at there is 500 channels. We have a larger machine, a supercomputer for 16, if you like just thousands tens of thousands of channels. And though the Stanford group were able to take one human sample, put it on not a tumor. I mean a patient, a patient, this is a whole genome, three billion letters read at least 30 times. So 100 billion data points. And then in 2.5 hours are able to sequence why read 30 times. What does that do? So that gives you enough coverage to rub out any errors. It's like 30 kids doing the spelling test. You get the perfect score rather than the brightest giving you a good score or the dumbest giving you not so good school. So that takes typically weeks to do, we did that in 2.5 hours. And that was actually 60 x coverage. So that's actually one genome every 90 seconds Now, less than 20 years ago. 2003. A global consortium took 10 years And to do one human genome and it costs three billion That's 90 seconds to sequence one. And that real time nature meant from sample to answer in seven hours. They were able to take a 90 day year old newborn who had severe epilepsy. The the clinical panel of care because it's only looking at known mutations. So you're not getting the whole picture. We were able to read the whole region around epilepsy and we produced a particular set of mutations were schoen and a drug regimen was given. Two weeks later the panel came back inconclusive life saved. And of the 12 patients they looked at in critical care in real time. Same day sequencing five of them got one of them actually got a heart transplant. Certainly remarkable. I mean, you know, both ends. One is you have something that can democratize care. So every center in the country clinic in the country can now sequence D. N. A. Instead of sending it out. And then you've got something where almost in real time we can sequence whole genomes to know what's going on and be able to pinpoint care. So this is you know, this is the beginning of truly personalized medicine. It's kind of exciting. I mean 20 years ago that genome was mapped and it usually takes about 20 years where we go up to the optimistic phase and then down into the valley of despair and I think it really is coming of age. And there are two important things, you know, genomics has been around and they've made great strides in understanding medicine and streaming data and helping inpatient healthcare. But it's been limited and now is the moment where making it affordable accessible and in real time so that patients are getting data immediately. That's when it changes everything because then this pact is made between the patient and the clinician who suddenly change and trust this new technology and that hasn't really happened in genomics yet. It's too abstract and too far away. That's a good point. So how much does this device cost? So this is $1000 and it's designed to allow a grad student to run a school project. And that idea that Everything just the hardware is $1,000 and then the cartridges, The cartridge to cartridge is $1,000. So we give the hardware away for free because we know that that democratization will change all the different things we can do on understanding the underpinning genomics of disease and patient care and by making it affordable it will just result in because it's such a simple thing. It literally is a strand being gone through a very small hole. We can make the instrument very cheaply and therefore have a model where boxes free. So I've seen this technology for over a decade evolved and I have to say I am wild and it is remarkable the company itself, you guys went public on the London exchange and how is it doing? I mean where is the revenue gonna grow? What's going to happen with the company over the next year or two. So 10 years in the making to make the technology work from academic inception And then five years ago we launched a product and we sell into over 85, well 100 countries now and into the research markets and that's our bread and butter. But where we are now after actually six years of commercialization and improvement and enhancements that you see with all disruptive technologies, think mobile phone, I think digital photography. Um so we are at a point now we are ready for prime time. We we have clinical accuracy and game changing time to result in critical care already well proven and you know turning heads but many, many more cases can be served and we're at that sort of inflection point but we are at the foothills with regard to diagnostics because the next step is the regulation and the verification but we have a lot of really exciting projects we're going to be working on. You know, one of our collaborators. Um it was a little bit crazy but what he did was he took your device and he hooked it to a dangle to literally go from his this device two starlink. So the notion that anywhere on the globe he can sequence and that data can go to the cloud be used by a country be used for policy, be used by a clinician. It's truly transformative. Yeah. And I think we think beyond the pandemic and we you know, I see this vision of the world where there will be a network of surveillance tools, This will be as ubiquitous as we have the sniffer machines in in airport security where we will pick things up. You see something in somebody talks about going from Nebraska to Beijing. Well you pick that up real time and there will be a global alliance because it's there are no borders here to protect and you know the chinese, the americans will all be one happy scientific community and you'll snuff it out because the way you stop a pandemic, the way you stop a fire is you put it out before it gets out of control. And and we really believe in that vision and the pandemic has really catalyzed that there's some amazing use cases across Siri Lanka across Japan where they're putting in networks across the whole island and we hope to be seeing some of that in Indonesia, that's pretty amazing. So your background is you're an engineer, chemical chemist, a chemical engineer. And now you're ceo of a large technology company in genomics is that difficult that transition? I do feel like an impostor quite a lot. Every time you ask me a difficult genetic question, I have to go and look it up. But I think in some respects as a disrupter, you actually don't want to know too much about how it actually works. You want to come from a different field. So being a physical chemist, being a chemical engineer. I no, there is incredibly smart people, you know the genetics of what we do. I have my written down bits that I think I understand and I stick with those. So sequencing DNA or RNA are obviously critically important but you know on top of D. N. A. Are what we call epigenetic changes. And the whole epic genome is a new concept is not new but we're slowly learning its relative and it's important and it will correlate to human uh physiology. Human traits. Can your system look at that? Absolutely. And it looks at it directly. So we saw that native strand going down the hole so that's we we do that and that is very unique to oxygen and a pool. What that means is as methylation the chemical corruption of the sea bass which results from the environmental impact of how we live and what we do in our environment. We can read that directly and that is absolutely game changing. And there are early publications really showing how methylation is important in neuro degeneration in cancer in pretty much every disease. And people have known that for 10 15 years. But to sort of quote Calvin if you can't measure it you cannot understand it. That's gonna be pretty cool. I mean as this can ubiquitously we can start to see what does it mean because we know a tiny bit of what's going on. You know when when we sequenced DNA classically we cut it into small pieces and we sequenced each of those and then we have a lot of computing power to put it together. Do you guys do the same or can you do the long story? We can do very long tracks which become really important, particularly in neuro degeneration where huge chunks like thousands of bases get cut and put in the wrong place and nature does that and they're really in critical, critical. So in the context of the small short reads 10% of the human genomes unmapped. We close that up and that will really help to change some of these really intractable areas of disease. Well as we finish up, I mean we have here we've sequenced These 16 genes that are related to lung cancer. Um and we're actually sequencing them over and over again as error goes as symptomatically to zero. Um and I want to thank you. I want to thank you because new technology is certainly critically needed in the world and especially giving access two people and clinics all over the world, not just the major academic universities. You've created something that could democratize one of the most important tools in medicine which is looking at D. N. A. And so thank you and all of your colleagues for doing that and we appreciate you being here in California. Thank you and thank you for your time