Do we now have near real-time gene sequencing? And if so … what does that unlock?
In this edition of TechFirst with John Koetsier we chat with Dr Roel Wuyts, principal scientist at IMEC and a professor at KU Leuven about gene sequencing, which used to take a lot of time.
Remember the Human Genome project? It started October 1, 1990 and completed in April 2003.
Now there’s a way of sequencing a whole genome in just 10 minutes for some sequences and a few hours for a whole human genome, which should unlock major new capabilities like personalized medicine and smarter treatment of currently deadly diseases.
Scroll down for full audio, video, and a transcript of our conversation …
Also see my Forbes story based on this interview …
Subscribe to TechFirst with John Koetsier
Watch: Real-time genome sequencing with wearables just years away
Subscribe to my YouTube channel so you’ll get notified when I go live with future guests, or see the videos later.
Read: Wearable DNA sequencing devices just 4-5 years away
(This transcript has been lightly edited for clarity).
John Koetsier: Do we now have near real-time gene sequencing? And if so, what does that unlock? Welcome to TechFirst with John Koetsier.
Gene sequencing used to take a lot of time, and I mean … a lot. The Human Genome Project, after all, which most of us know about, started October 1st, 1990 and completed in April of 2003. Now there’s a way of sequencing a whole genome in just hours, which should unlock major new capabilities.
To learn more, we’re talking with Dr. Roel Wuyts from IMEC, a major global research hub. Welcome, Roel!
Roel Wuyts: Hi John. Thanks for the invitation.
John Koetsier: Hey, super happy to have you with us. Tell us about this new DNA analysis platform that you’ve released.
Roel Wuyts: So we are software people, and so we optimize software pipelines and we focused here on DNA sequencing.
And so we’ve done that for quite a while and we released the latest version which can now do a 50X coverage whole genome — that’s one that’s quite often used — we can now do that in under 10 minutes for roughly speaking $1.
John Koetsier: Wow! That is significantly fast and significantly cheap. How much faster is that then typically what you would have done in the past?
Let’s say, you know, before this process you’re a scientist, you’re in the medical field, you needed a genome sequenced. Where do you go? How long does it take? How much does it cost?
Roel Wuyts: Good question. So, in hospitals they use de facto standards in many places called GATK, which is from the Broad Institute — so, [a] well-respected tool that gives very good outputs, which is what we want.
And so that’s a pool for this whole genome, 50X a whole genome would take roughly [an] order of magnitude of 6 hours and cost $20 if you would run that in Amazon — that’s the elPrep one.
John Koetsier: Yes.
Roel Wuyts: The standard one, which is GATK tool, that would take between 2 days and 4 days. So 4 days, if you take the classical GATK tool, which is the output that most hospitals still prefer. There is an optimized version where Intel actually optimized part of the process, but that still takes 2 days.
John Koetsier: Wow.
Roel Wuyts: So that’s much better than 4 days, but without prep we are done in, let’s say, yeah, 6 hours. So that’s still a lot faster with exactly the same outputs than the 4 days version that hospitals prefer.
John Koetsier: Interesting. So you mentioned here a number of hours, and then in the first question you mentioned something like 10 minutes or something. What’s the difference there? And what makes up for the extra speed?
Roel Wuyts: Yeah, the thing is, so you have several analyses that you would like to do. So sometimes you take the whole genome and then it would take, roughly speaking, these are the 6 hours. Sometimes you take an exome, which is like a smaller part, and then that one you can do in 10 minutes.
So it depends on the size and the analysis you want to do.
John Koetsier: Now, what kind of genome are we talking about here? Are we talking human genome? We talking a genome of a fruit fly? Which one are we talking about?
Roel Wuyts: The human genome. So, the ones that most hospitals prefer.
John Koetsier: Haha, yes, I guess they don’t need the genome of a fruit fly do they?
Roel Wuyts: I’m not sure. I don’t want to … it could be very interesting research there. So …
John Koetsier: You’re a cautious scientist. I get it.
Roel Wuyts: Yup.
John Koetsier: So, tell us … I mean, speed is better. Cheaper is better … obviously, just in and of themselves. But what does this unlock? What does it do? What can we do now that we couldn’t do before?
Roel Wuyts: Yeah, what we are helping enable is the move from using DNA and DNA sequencing based tools from research into a clinical setting … and a clinical setting in the daily practice. So, most hospitals do clinical research as well, and their speed and cost is also important, but the volumes are much lower. But if we talk about a hospital that buys a sequencing machine for daily clinical practice, they really want to use this.
And if you think about recent DNA machines that you use at volume, they can produce up to 90 about — sorry, 9,000 genomes a year. And if you need to do that, if you really will let this machine run for a full year, then the processing of the outputs would take a year and a half to complete.
And that’s a problem, right? It’s not very cost effective, and so that’s what we’re trying to enable. So help hospitals take that step to use DNA in a daily setting, cost-effectively.
John Koetsier: That’s pretty interesting. There used to be a saying in the computer industry that “what Intel giveth, Microsoft taketh away.” [laughter] And I’m glad to see that the software is starting to give back here.
So, let’s dig into that just for a little bit, because what you’re talking about is medicine and medical treatment that is tailored to somebody based on their genome, tailored to them based on what genes they have. So I assume there’s some real basics in there, right? Do you have the genome pattern for a particular disease or predisposition to some particular thing or not, right?
But are you also working into higher order things like tailored medication, that sort of thing?
Roel Wuyts: That’s a great question, and that’s indeed one of the possible directions that we want to enable. But before we can do that, if you want to go to tailored medicine, then you need to have quite a large dataset.
And so, by making now this initial genome sequencing practical and cost-effective, this pool of available data will become bigger. Since if my DNA is sequenced and I’m 1-in-10, there’s only 10 other people then you can compare it with 9 other people, but chances are that I will be a lot different.
John Koetsier: Mm-hmm.
Roel Wuyts: But if we can do that for millions of people, then the chance that I resemble somebody for which we have more information is much bigger. So that’s indeed one of the subsequent goals that we want to enable to go to such … those type of research and then linked to care.
John Koetsier: Interesting.
Roel Wuyts: Personalized care.
John Koetsier: So, you’re building the software platform that’s making the sequencing much, much quicker. Are you also building a platform then to enable what you’re talking about here, which is that, hey, if you’ve got my genome, you’ve got your genome, you’ve got several million others — you can make more intelligent inferences about how to treat somebody.
Are you making a software platform to enable that, or is that somebody else’s job?
Roel Wuyts: No, we’re also working on that in collaboration actually with other partners. So we worked on that in combination with a University in Belgium, the KU Leuven. So we’re collaborating to make that possible in a privacy-preserving way, because that’s important.
We’re talking about the genome of people. So that’s not something that you just want to pool in a single large database and let just anybody do queries, right? So you want to enable research and be able to conceptually pull this data, but without actually pulling it. And so that’s a type of distributed querying and distributed analytics that we are engaged in.
John Koetsier: Very interesting topic. Of course, that’s a big deal lately with sites like Ancestry, or 23andMe, or services like that, right? Like who owns DNA? Who owns your DNA record and what can they do with that? And who has access to that?
Some big questions there which are challenging, but obviously you’re doing it in some sort of differential privacy or distributed way so that it’s not releasing your personal DNA and tagged to you as well.
Roel Wuyts: Yeah, exactly. So, and also subsequent effects, say if you let people query it even in a safe way, you don’t want that indirectly it still leaks information.
John Koetsier: Yes.
Roel Wuyts: So, it’s a very interesting area because there’s a lot of data. It’s very complex analysis and you want to make sure that it preserves privacy that’s ethically correct.
John Koetsier: Very, very interesting. Okay, so how do scientists get access to your innovation? How do they start sequencing genomes 10X faster?
Roel Wuyts: Well, it’s luckily not that hard, so we have open sourced it. So elPrep 4 — that’s a predecessor — has been on GitHub for quite a while now. The new version elPrep 5 that we released will be there not too long from now. And we’re actually releasing it with a dual license.
So it’s open source license on the one hand, that’s AGPL, so people are forced to potentially give back useful extensions they make … that’s what we want. But for people for which this is too restrictive, they can also get, we called it — or we will call it — a “commercial license,” so one that you are not forced to give back the extensions you make, but then you need a paying license, because then we can use that money to again improve the open source version.
John Koetsier: I like it a lot. You can pay by either helping improve the code, or you can pay by forking over some dollars or euros.
Roel Wuyts: Yep, that’s the intention that we get going because we want to continue working on genomics software.
John Koetsier: Excellent. Excellent. Now let’s extend this out a few years, and we talked about this a little bit as we were prepping for the show.
I’ve got a computer on my wrist right now — many of us do — it has a lot of sensors. Do you foresee a time when you might be able to … somebody might be able to sequence a genome in their basement, maybe on a wearable or something like that?
Roel Wuyts: Yeah, that’s indeed a good question because, so IMEC … it’s actually most known for all the work in nanotechnology and applications of nanotechnology. And so we have groups working on various aspects for making such a device a reality.
And we actually have a lot of building blocks already.
So I have colleagues that worked on microfluidics, and that means that you could enter a saliva sample or a blood sample directly on such a device. So without having to go through complex preparation steps that would then be done with this microfluidics in this device. We also have chips sensors that could do the sequencing itself on the device. And we then have all the knowledge and the analysis — that’s what we are reached now with elPrep.
But we could scale that down and put that same compute in such a small device by leveraging some of the work that IMEC is doing on a very small tiny computer. So, bundling all of these building blocks, we have done some internal brainstorming and we think within the 4-5 years time frame that’s something that we could actually do.
So it’s only a question of integrating these building blocks. That’s still challenging, but still a lot of the basic blocks are actually there. So … then you would have a wearable where we think you enter directly a small amount of blood and within 4 hours we could have your fully sequenced DNA and some analysis.
John Koetsier: Very interesting. Just imagine how scientists could use something like that if they’re in the field and sequence genome of a new-found species almost immediately. As well, of course, of the medical benefits potentially of having genome sequenced of a hundred million people, a billion people, or something like that …
Roel Wuyts: Yeah.
John Koetsier: … what it could unlock for medicine.
Roel Wuyts: Yeah, there is a lot of very interesting applications we could get. So we were thinking about what, indeed what can we do. For example, to give an example, in oncology, there is a number of cancers that currently we can’t really treat, [unclear] myeloma. So, particular types of blood cancer.
But they evolve very fast, and so at this point, it is therefore hard to keep track and to tailor the medication to, let’s say, the current evolution of this cancer that a patient has.
But with this tool, we could, for example, a bit like diabetes, you could like give a daily amount of blood that analyzes the DNA and that then can adjust the medication. And we think that if you could do that, that you could turn this much more into essentially a chronic disease rather than one that currently is fatal … much too early, for example.
John Koetsier: Wow. Very, very interesting. Well, Roel, just want to thank you so much for taking a little bit of time — I know it’s late in your evening — to spend with us, and chat with us, and tell us what’s new.
Roel Wuyts: You’re welcome. I was very happy to be on the show, the podcast.
John Koetsier: Excellent. Well for everybody else, thank you for joining us on TechFirst. My name of course is John Koetsier. I appreciate you being along. You’ll be able to get a full transcript of this within a few days at JohnKoetsier.com. The story at Forbes will show up after that. Plus, the full video is available on my YouTube channel.
Thank you for joining. Maybe share with a friend. Until next time … this is John Koetsier with TechFirst.