In addition to that, you may also raise your virtual hand to address your questions verbally. For participants joining via phone, to raise your hand, use star 9 on your phone's dial pad. When you then get selected to ask your questions, please follow the instructions from the phone and press star 6 to unmute yourself. One last remark: if you'd like to follow the presented slides on your end as well, please feel free to go to roche.com/investors to download the presentation. At this time, it's our pleasure to introduce you to Bruno Eschli, Head of Investor Relations. Bruno, the stage is yours.
Thanks, Henrik. And could I have the first slide, please? So welcome to our second IARC call in 2025, focusing on our revolutionary SBX nanopore sequencing technology. And let me quickly take you through today's agenda. Overall, we have 60 minutes for today's call, and we want to give you the opportunity to ask as many questions as possible. But before we start with the Q&A, we thought it would be helpful to have a quick wrap-up presentation to summarize what has been presented earlier today at the customer webinar. And also, please be aware that you can find links on the IARC homepage for today's events, the IARC event, as well as the customer webinar. So the introduction presentations now will be given by Matt Sause, our CEO of Roche Diagnostics, and by Josh Lauer, our Global Head of Customer Area Molecular Labs.
Thereafter, we will open the Q&A session. For the Q&A session, we will also be joined by Palani Kumaresan, our Global Head of Roche Diagnostics Solutions. The Q&A session will be moderated by IARC Officer Birgit Masjost, who is in the lead for diagnostics within the Roche IARC team. And let me also use this occasion already to thank the IARC team members who contributed to preparing today's call, which are Andreas Lins, Anita Tang, and Jörn Bayer. Could I have the next slide, please? So I don't want to steal too much time here, so only one quick reminder: the next IARC call is scheduled for April 4th and will cover the latest clinical news flow in neurology.
This will include the two-year Elevidys EMBARK data in DMD to be presented at MDA, and the Prasinezumab phase II PADOVA data in Parkinson's disease, as well as the latest Trontinemab Brain Shuttle data in Alzheimer's disease, which are to be presented at AD/PD. And with that, I hand over to Matt to kick us off with today's SBX nanopore technology unveil. Matt, please.
Thanks a lot, Bruno. So today really marks an important moment in the history of Roche Diagnostics, and following on the webinar that I'm sure many of you attended, which featured Mark Kokoris, the inventor of the SBX technology, as a speaker, I'd maybe want to start by thanking Mark and the entire Roche sequencing team for getting us to this important point. So here, I'd like to provide a brief overview of why we are here today and why we're so excited about the potential of the SBX technology. So first, taking a step back and talking about the importance of the next-generation sequencing market, so this is currently valued at approximately $6 billion USD, and it's projected to grow at 9% a year, and this is really driven by two segments: biopharma and clinical.
And I would call out that the clinical segment is growing at double digits, really driven by the broad uptake of applications such as therapy selection and minimal residual disease monitoring for oncology, where NGS is really becoming the standard of care. These are areas of high interest to Roche, and we see these applications as future growth drivers for our diagnostics business at large. So entering this space, the Roche SBX technology has the potential to deliver high accuracy with unprecedented throughput, speed, and flexibility. Roche is uniquely positioned to deliver differentiated solutions spanning both research and clinical segments by leveraging our broad organizational competencies. And I think that's one of the unique strengths of our organization. Now I'm going to hand it over to Josh Lauer, the Head of Roche Molecular Labs, to take you through the SBX technology in a bit more detail.
Over to you, Josh.
Thanks, Matt. Roche's interest in sequencing is not new. In fact, we offer a wide range of solutions across the sequencing ecosystem, including library prep, automation, assays, and interpretation software. Of course, in the absence of our own platform, to date, these solutions have been focused on third-party sequencers. So if you go to the next slide, that is why we're so excited to introduce SBX today, the technology behind our sequencing platform in development. Now, as Palani mentioned in the webinar, it's important to note two things. One, this platform will be open, so researchers and the sequencing community can use it to build their own solutions. At the same time, we will have it as part of an end-to-end sequencing ecosystem that we can bring into a clinical setting over time along with our assays. Next slide, please.
So if you didn't tune into the webinar and Mark Kokoris' presentation, a quick summary of SBX. It's the combination of two innovations. The first is our sequencing by expansion chemistry, which came from Stratos. This turns DNA into a longer molecule and significantly improves signal-to-noise. The second is a sensor-based module. It's Genia CMOS technology. This enables precise, simultaneous, high-speed measurement across eight million wells. Together, this combination has a unique ability to generate accurate reads at incredible speeds with flexible throughput. Next slide, please. Specifically, what are some of the benefits we see with SBX? Today, customers of all sizes are forced to batch samples in order to run in a cost-efficient manner. SBX has the potential to upend this dynamic in the market with a reusable sensor module that minimizes the cost penalty to run fewer samples.
We also have flexible read lengths with a simplex mode for longer reads and a duplex mode for traditional short read applications. I'll say a bit more about these modes in subsequent slides. Now, in initial demonstrations, SBX has shown the potential for high accuracy, Q39 duplex-based quality, and variant calling performance comparable to commonly used technology. Because SBX is incredibly fast, we've also demonstrated the potential to sequence a genome from sample to result in less than seven hours. The same speed means we have high throughput, and we've demonstrated as high as seven genomes per hour. Last but not least, the reusable sensor module also means that SBX can be very cost-efficient. Next slide, please. So we are in the process of developing a high-throughput sequencing solution based on SBX technology.
The solution will have two instruments: a synthesis instrument that turns DNA into expanders and a sequencer that turns those expanders into base calls. Now, we are keeping these instruments separate for a reason, namely to maximize flexibility. So if you advance, I think, two slides, we can see how expander synthesis works. This is an automated instrument that can process up to four library pools at once. That gives it the ability to support flexible throughput and also to feed more than one sequencer. Now, a highly engineered polymerase transforms DNA into a synthetic molecule. This new molecule contains reporter codes that correspond to the original DNA bases, but they are significantly larger, so only one base reporter fits in the pore at a time. If you advance to the sequencing slide, one more.
Sequencing is performed by running these expanders through 8 million pores in parallel on our sensor module. I'd like to call out two fundamental aspects of this process. The first is that expanders are moved through pores in a very controlled fashion. Voltage pulses are applied to advance the molecule one reporter code at a time, and this generates a clear, discrete signal. The second very important thing is that the entire sensor module can be reused, and this makes it more cost-efficient. Next slide, please. So back to my earlier reference, customers can run SBX in two different modes depending on their application. If you go forward one more slide, we'll see that we have a duplex mode that links both strands of DNA into a single read for high intramolecular consensus accuracy.
This mode supports inserts in the range of 200-300 base pairs, and it's great for applications like somatic or germline sequencing. We also have a simplex mode, which runs a wider range of read lengths at unparalleled ultra-high throughput that we've really not seen in the industry before. So advancing slide forward, we wanted to demonstrate the potential of duplex mode. So we ran a standard set of reference genomes from the Genome in a Bottle Consortium. As you can see from the numbers on the slide, the level of accuracy is very high. It's consistent across samples and also across multiple variant calling methods. Specifically, with SNV and indel accuracy well above 99% and duplex Q-scores of 39, we believe SBX is demonstrating the level of accuracy needed for clinical applications. Next slide, please. One specific application is neonatal intensive care unit sequencing.
We've said that SBX is fast, so we wanted to see how quickly we could sequence a single human genome. With a bit of work, the team came up with a protocol for this setting, and working with the Broad demonstrated it could be done in under 7 hours. Now, others have demonstrated a fast genome in very special situations. But SBX has a unique potential to make world record speeds routine and cost-efficient. If you go forward one more slide, we wanted to also demonstrate simplex applications. Here on the slide, you see a couple of experiments we did with the Broad Institute that demonstrate the high-throughput potential of this mode. The experiment on the left is a multi-omic Perturb-seq experiment, which generated over 30 billion reads in only four hours of sequencing. This is ultra-high throughput. The experiment on the right is a typical single-cell gene expression experiment.
However, what's unique about this is that it demonstrated SBX's ability to generate longer reads. In one hour of sequencing, we yielded more than 200 million reads, 1,000 base pairs or longer. These reads are useful to detect novel isoforms that can reveal gene function or disease mechanisms. And typically, researchers doing these experiments require more expensive long-read technologies, and SBX has the potential to make this type of research much more efficient. Next slide, please. Obviously, SBX is generating raw data at unparalleled high rates. Because of this, Roche is developing a GPU-based accelerated compute module to provide near-real-time base calling. This itself is another area of innovation, and over time, we intend to provide additional on-instrument functionality. We also recognize that customers like to run their own pipelines, bioinformatics, so we intend to make SBX tools open-source for the community to leverage as they see fit.
Next slide, please. So we're very excited about the potential of SBX. This is a general-purpose sequencing technology, and it can accelerate a wide range of research, translational, and eventually clinical applications. The throughput, speed, and flexibility of SBX are helpful across the board of sequencing applications. We see unique potential for low-volume customers who traditionally have been unable to run small batches without a significant cost penalty. And we also see unique potential in applications where time is of the essence, such as the NICU or oncology. Next slide. As a group, Roche is uniquely positioned to maximize the potential of this technology. Not only will it form the core of diagnostic sequencing product development, it can also be used to accelerate drug discovery and our personalized healthcare business with pharma. We would like to see clinical labs such as Foundation Medicine and others build applications on the platform.
To support this ecosystem, we will use a combination of Roche tools and third-party solutions. SBX will be an open platform for the entire sequencing community, and we look forward to seeing what they'll do with it. So with that, I'll turn it back over to Matt.
Thanks a lot, Josh. So now I'd like to thank the entire team for the work in getting us to this point with the SBX technology. And so now I'd like to really just emphasize the key takeaways here. First, the NGS market is highly attractive and is growing at 9% a year, driven by the clinical and biopharma segments. Second, the Roche SBX technology represents a step forward for NGS and has the potential to deliver high accuracy with unprecedented high throughput, speed, and flexibility. Third, the SBX technology has the potential to enable and also accelerate both research and clinical applications. And fourth, we are building an open platform, which will be the core of our end-to-end sequencing offering, which we are strongly positioned to populate with relevant content. So in closing, the unveiling of our SBX technology is the culmination of years of work.
It demonstrates our commitment and capability to deliver innovation that has the potential to change the field of sequencing. It also represents a tangible example of our ambition to continue as the world leader for in vitro diagnostics, as we laid out in our investor day in May of last year. So next slide, please. So I'd also like to conclude by mentioning two upcoming events. And the first, as you heard, we will hold a workshop at the AGBT conference on our SBX technology, highlighting some of our earlier customer data that you may have seen on our webinar. Second, I would like to remind you about our Roche Diagnostics Investor Day, a hybrid virtual and in-person event in London on May 27th. Here, we will highlight our SBX solution, but also our exciting pipeline across all of Roche Diagnostics. So I look forward to seeing you there.
And so with that, I'll hand it back over to Birgit.
Oh, and a very warm welcome from my side as well. We're happy to take your questions now. If you would like to ask a question, please raise your hand or type the question into the Q&A tool of Zoom. So the first question goes to Dan Brennan from TD Cowen.
Great. Thanks, Birgit. Thank you very much. Can you hear me okay?
Yeah.
Terrific. Thanks. I was wondering if you could start. I know there's nothing on price yet, but you talk about this reusable kind of flow cell, I guess. Is there any way to put it in context like most of the players will quote price per genome or price per gigabase? Is there any way to think about it if you're not going to give numbers yet? Any way to think about, relatively speaking, in the zip code below above, just given the profile that you have? It was the first question. And then the second question was just back on accuracy. So the simplest accuracy is what, Q20? Is that right? And I know you showed duplex at Q39, but I was just wondering if you can expand upon that a little bit, given it was unclear. I know 50% of the reads match on duplex, 50% don't.
So I just wanted to understand, is that Q39, you run duplex mode and that's what you're going to get Q39, or is there some ratio that you're calculating? Thank you.
Josh, you want to take the first one, and I could take the second one if you want.
Sure. So thanks, Dan. As you noted, we are not disclosing pricing information at this time, but I guess what I could say is that obviously we are trying to design a system that is priced competitively and can meet market expectations. We're making a lot out of flexibility, but this is a high-throughput system. Let there be no mistake. We intend to be competitive in the high-throughput sequencing market. But with the flexibility of SBX, we also want to make efficient sequencing accessible to lower-volume customers as well. I'll let Palani go on the other question.
Absolutely. I mean, I can quickly jump on the second one. So the way to think about our accuracy is also by application. If you're thinking about oncological applications where we will use our duplex sequencing mode where we are connecting the sense and the anti-sense of the same molecule, in that case, the average accuracy is Q39. And in terms of when you think about mapability, which is what really matters in these applications where we talk about our F1 scores, there you can see for our SNVs and indels, we are really comparing very nicely to commonly used technology. So I would say there is no hybrid mix going on or any of that. I mean, you should think about it as Q39 is the accuracy, is the duplex accuracy for those applications.
But then there are other applications like counting applications that Mark also shared during the webinar where you don't need as high accuracy, but you're really looking for very high throughput. There you can see for, I mean, for a linear simplex, the accuracy is 99.3%, as Mark was sharing, the raw read accuracy. And that's good enough for those counting applications where you're really looking for very ultra-high throughput, which is one of the unique features of our platform as well.
Great. Thank you.
Hey, great. So the next question goes to Matthew Weston from UBS.
Thank you very much. Hopefully, you can hear me.
Yes, we can.
Two questions for me, please. And I guess I'm going to call them simple pharma questions. Matt, previously, when you set out the potential for Mass Spec, you were prepared to give a target revenue for what you hoped could be achieved in 2030. I wonder if you're prepared to set out something similar for next-gen sequencing. And then I'll come with a second question in a minute.
Thanks, Matt. I mean, we talked about some of the key applications that are driving next-generation sequencing, right? So for example, minimal residual disease, the therapy selection like what we have with Foundation Medicine. These are applications that are growing. If you look at MRD, 60% a year. You look at therapeutic selection, you look at about 10% a year. So the actual market share that we are planning to be able to disclose will come a little bit closer to launch, and I think at the moment, this was the tech reveal. And I think once we have the final product configuration, I think we'll have a good conversation about how we see our share developing. I don't know, Palani, do you have anything you'd want to add to that?
Okay. Understood. And then the second question, if I could.
But I guess what I want to say there is we're going after those applications that are high growth. And if you look at that, we want to be a player in those. But in terms of our overall ambition, I think once we've really got that initial product launch configuration, something that we put in front of you, then we can talk about that piece.
Understood. Totally understood, and then the second question, if I think of the CGM reveal, basically what happened or what became clear was there was a bottleneck really for sensor manufacturing. So we had the launch, but then we had a period of limited supply. When we get to the launch and if we see the industry uptake that maybe early feedback suggests, do you think you're actually capable of scaling the technology to meet customer demand at the start?
Absolutely. I mean, I'll simply answer by saying yes. I mean, that's very much on our minds. And we are also very systematically, as we mentioned right now, we are in early access. So we are systematically kind of engaging to also see what that customer appetite is right now. And as we go into our launch year, which is 2026, we are also very much systematically looking. And as you could see, Mark Kokoris, he also showed multiple of our sites and manufacturing facilities are deeply engaged in this, precisely with the idea of the appropriate scale-up in mind.
Yeah. And we've been looking at some of those pilot transfers with our facilities in Penzberg for a significant amount of time. I think what we saw with CGM is we did quite a number of things in parallel to speed our time to market. And that prevented us from being able to do some of that sensor manufacturing in advance until we had that final configuration. But I think with the chemistry on SBX, that we've been able to start those transfers a bit earlier. So fully agree with what Palani's saying.
Many thanks indeed.
Yeah. Thank you.
Okay. Cool. So next question goes to Justin Cooper from Nephron Research.
Hey, guys. This is Justin on for Jack Meehan, and thanks for taking the question. Just wondering, so at your Analyst Day last May, you mentioned a path to transition the Foundation Medicine volumes over to the new system. Just wondering what's a realistic timeline under which you think that could happen. Thanks.
Josh, you want to take this one?
Yeah. I don't think we're ready to disclose any timing on that. As mentioned earlier, the system isn't even launched yet. We're in early access. So I think as we get closer to product launch, I think we'll have a better sense for what we think the customer base and utilization will look like in those very early moments after launch. And so I think if we wanted to talk about those roadmaps, hopefully we'd have a better sense of it then. It's worth noting that Foundation Medicine is an independent lab that makes their own product development decisions. And so this is something that I think they would have to consider and speak to over time.
Thanks.
Okay. Great. So the next question goes to Subbu Nambi from Guggenheim.
Yes. Thank you, Birgit. Josh, great to meet you. And great presentation, Matt and Palani. How would multiplexing of samples work on the system, one, and second, is there a requirement for local computing for data processing? And if there is, how would this compute relative to Illumina's onboard driving for clinical applications?
So I can take those. And Palani, feel free to follow on if need be. So multiplexing is rather straightforward. Like any other technology, we have sample barcodes that are part of a library prep process. And so demux is one of those compute steps to your second question that we'll allude to. So if that's sufficient for the first answer. So the second one, I want to be clear. We are developing accelerated compute that goes with the system. So this is part of the system. And as John Mannion presented during the webinar, the intention is for that accelerated compute to provide near real-time base calling and actually also demux, so intramolecular consensus as well for duplex reads. So hopefully that gives you a better sense that accelerated compute is in scope of the system and the scope of what that accelerated compute will do.
So Josh, if I can just summarize for my own, for simplicity's sake, one, you can multiplex like the way you're multiplexing on existing sequencing system. And second, there is no special compute requirement. It's probably more or less onboard compute is what you are trying to do as well. Did I get that right?
That is correct. And for secondary analysis, like any, with any current technologies, you can do it on the cloud or on-prem, depending on the customer's preference.
Thank you, guys.
Okay. The next question goes to Simon Baker from Redburn.
Thank you for taking my questions. Three quick ones, two of which are kind of connected. I just wonder, given the profile of the product, could you give us an idea of what the typical customer looks like? Who are you going after here? And if there were multiple customer types, that would be great to give us some sort of color. And related to that, if you're lowering effectively the barriers to entry and making this easier for people and allowing people who are not currently using NGS to get in here, what do you think you could do to that 9% growth rate going forward? And the final question, just on IP, I wonder if you could give us an idea of where you stand there. It looks like the expansion protection is to the late 2020s, the detection to the mid-2030s.
But some idea of where we stand on the protection would be very helpful. Thank you.
Josh, you want to take the first two maybe, and then maybe the third one I can take.
Perfect. So the first question, sorry, I'm losing track. The second one was what we could do to the 9% growth rate. The first one was what segments we're going after, correct?
Correct. What does a customer ?
What the target customer for this technology looks like.
Yeah. Look, I apologize in advance for what will sound like a vague answer. But the reality is, as I mentioned in the presentation, this is a general-purpose sequencing technology. Now, we're not saying it's a long-read technology. We're saying it's short and longer reads. So if you wanted to take a first pass at what portion of the sequencing market we're aiming at, I would start there. But within that space, sequencing is general-purpose. This is a general-purpose sequencer. And we've tried to be very clear that we are not only looking at clinical applications, but that we will make this an open research-use-only platform for the community. So that's a pretty broad base to start from. As Matt mentioned, of course, we are a clinical diagnostics company. So our long-term interest is to focus on those clinical applications. But we're starting with research use only.
We don't want to limit ourselves to only the clinical applications, which will come over time.
Yeah. And Josh, like you and Palani were saying, for some of those counting applications with the ultra-high throughput, there's tremendous applicability. But then the Q39 accuracy could really fit well with things like minimal residual disease detection. So it's really got a broad potential. And I think we don't want to sell ourselves short with the kind of customers we could engage with.
And the second one, Josh.
And then the second question was.
Go ahead.
Yeah. So that's a good question. I have to admit, I'm not sure how to think about this, but my first blush answer would be to say that obviously, anytime you can unlock barriers in the industry, you expect, whether we're talking Jevons Paradox or anything like that, is they talk about an AI unlocking more demand. Of course, that would be our aim, right? We are a broad diagnostic company doing billions of tests a year. And so we would love to see sequencing going all over the place. Could we pull forward some of that growth? Yes. We would like to think that SBX could be a step function change in the market. But we have not launched the product yet. We are very excited about the potential of what this technology can do.
I think lowering the cost for mid and low-throughput customers should advance the expansion of that market. I can't give you a number for the overall market growth at this point in time. No, I do think that we should be able to move the needle.
Good. No, thanks a lot, Josh. And for this third question, I'll keep it pretty brief. I mean, and I'll go back to a quote that Mark used at the very beginning of his webinar. He said, "One of the best ways elegantly to sequence a DNA molecule is not to sequence a DNA molecule." And that is one of the kind of the fundamental pillars of SBX technology. And what I would say is we have I mean, we do all of our due diligence on the IP side, and we are very thoroughly assisted. And we know that we are building a very nice family of IPs that will really protect our technology in years to come.
All right. Thanks so much.
Next question goes to Puneet Souda from Leerink.
Hi, guys. First of all, congrats to the team here on an incredible job and many years of work. So great to see that, Matt and Palani and Josh. So first, I mean, if I could just frame it, look, you have a high-performance system here. The speed of the system suggests that it can run far ahead of the current platforms in a shorter period of time. Does that mean can you talk a little bit about it? What does that mean for the install base? Because in the research realm, does that mean fewer installs, at least to start with, and then potentially expansion from there? And then on pricing, I know you are not providing that today, but maybe just help us understand when can you start to provide pricing or some directionality on that?
And then lastly, if I may, clinical market, your peer company that is in the market today, it took them many years to get established. It was a different time frame. Even with Foundation Medicine getting that established, getting into that account and NIPT and all that took many years. So market is much different today. So just trying to understand, how are you thinking about the speed into the clinical market, which is much different versus 10 years ago?
I'm happy to give you some. Guys, feel free to jump in as you see fit. So on the first one around how speed impacts placements, I would suggest that speed actually has two dimensions to it. One is just the faster you are, the more throughput you can get if you want to keep loading samples. So obviously, this is a high-throughput system. And I would think of it the way you would think of the introduction of any other high-throughput system. Like what does that do to placements and elasticity of demand and how many more samples do people want to run when they have access to higher capabilities? I think what we've seen through the history of the sequencing market is you keep giving high throughput and they keep doing more sequencing. So I think Illumina's trajectory historical is the best evidence for that.
But I would add what's unique about ours and different than Illumina's high-throughput systems is speed, also unlocks new applications that might have previously competed with other forms of diagnostics technologies because they wanted a fast turnaround, and traditionally, sequencing has been rather slow and with runs taking a day or two days, especially if you want good cost economics, so the ability to deliver economical sequencing very fast can open up new use cases and new applications, so that's one of the things that we think is highly differentiated and we're very excited about.
Josh, I wanted to maybe quickly add to the first before you go to the second. Also in research settings, we clearly see there are many very sequencing-hungry applications that people want to use. We see it within Roche itself. When we look at our gRED organization or pRED organizations on the pharma side, when you're looking at perturbations and you're looking at Perturb-seq or so on, these are extremely sequencing-hungry. So it's going to really unlock many of these applications where people were still not able to get to with the current paradigm of sequencing.
Right, or even like Josh said, like Mark showed some of the data from the Coriell Cell Lines and some of those rare diseases, and those could be done via PCR today, well, if you're able to do a whole genome sequence in a matter of hours, that it may be something that then moves to NGS. And I think that's something we're going to explore as the technology matures.
Josh, you want to go to the second? Yeah, please.
Yeah. So your second question was when we'll be able to reveal more information about pricing. The simple answer is closer to launch. We've talked about launch being in the 2026 time frame. So I don't have a date for you, but I guess I would say as we get closer to the end of the year and we move into 2026, that'll be the right time frame for us to be disclosing more about that. And I have to admit, I do not have paper and pen with me here. I have forgotten your third question.
Yeah. It's just the clinical market. I mean, it was a very different market 10 years ago. Yeah.
So I do think it's a very different market. I do think clinical always takes some time because you want to validate things if you're switching them, right? However, we are seeing a burgeoning and growing clinical space. And as long as you have highly accurate sequencing and you can show comparable results, it is becoming easier to substitute. You still have to do the work of showing, validating and showing clinical utility if it's new or just doing bridging studies. But this is not necessarily a new problem. We already see clinical labs converting over generations of Illumina sequencing devices. And we see them now starting to evaluate other technologies. And we think SBX will be a highly attractive technology for them to consider as they look at their own roadmaps going forward.
So that's not a time frame, but I don't think this will take nearly as long as it took the first time around.
Sorry, Josh, and just an anchoring comment as well. I mean, we are the world leader in laboratory-based clinical diagnostics. And with that comes our infrastructure around service, support, commercialization, and things that really customers look for when they're trying to pick a partner for their laboratory practice. I think that also can't be understated.
Also, I mean, our deep understanding of regulatory landscape that also very much comes into play here.
Yeah. And we're always in deep consultation with our laboratory partners on that topic.
Thank you, guys.
Good. Next question goes to Dan Leonard from UBS. Dan, you need to unmute yourself.
Sorry about that. I didn't see the pop-up. Thank you for taking the question and congratulations on the announcement. I'm curious, what can you share about the reproducibility and robustness of the instrument that you're already seeing in early development and what might be targets by launch?
Happy to take this. I would say because we're in early access, we are seeing good reproducibility. Obviously, we've done multiple runs at multiple sites with multiple instruments, and we feel very good about the stability of the technology. You're right that we have not quoted any figures in terms of uptime or meantime between repair visits or things or variations. That will come closer to launch as we have matured through our early access program and have even more data to finalize specs and the robustness of those specs. I do want to be clear that the data that we have shown today in the webinar are not the specs of the final platform. This is a demonstration of the power of SBX and what we know it can do when we've done real-life runs with the technology.
But ultimately, specs and the reproducibility and characterization of those will come closer to launch.
Thank you.
All right. Next question goes to Mike Ryskin from Bank of America.
Hey, can you hear me?
Yep.
Okay. Great. I got that part down. Josh, I want to pick up exactly kind of on what you said. I thought that was a good point, and something you talked about in the earlier webinar is that you highlighted a lot of the different opportunities you have for further improvements. So the way I kind of think about this is clearly you've been working on this technology for a long time given when you acquired Stratos, when you acquired Genia and everything that's been going on behind the scenes. So my interpretation of it is you're unveiling it now. You're launching it now. You see that you think that what you have now, the Q20, the Q39, the throughput, it's good enough for applications now, but it can continue to improve.
I'm not asking for where it's going to be in a year or two years from now, but just sort of can you give us any kind of roadmap or where you think the biggest opportunity or the biggest payoff could be, whether it is in accuracy or throughput or specific use cases? Just sort of help us envision the roadmap from early access to full launch.
Sure. So I think Mark alluded to this in his webinar presentation. I think late in the webinar, if you want to go back to it, he had a slide talking about directions for future improvement, but I'll call out a couple. One is that although this technology has been a long time in development, I believe Mark alluded to the fact that a lot of things have come together more recently, and so my perspective and the way I would suggest characterizing it is that we are still in the very early stages of what this technology can do. A lot of the historical work was in putting it together into what is really an impressive feat of molecular engineering to make this type of performance possible, so on one hand, we already have what we believe are very competitive specs.
We don't actually think any improvement is needed for this technology to be a viable competitor in the space once it's launched as a product. On the other hand, we are early in our optimization journey for where this technology and its configuration can go. So, we cannot commit to any roadmap of improvements, but obviously, Mark and team and the whole sequencing unit will continue to look at optimization along a variety of dimensions. Some obvious ones are, can you lengthen reads? Can you boost accuracy? But also, can you boost throughput? We have not hit the limits of the throughput of what this technology can do. Mark alluded to a few mechanisms that you can use to drive throughput. But summing it up, impressive performance characteristics as we see, and headroom along multiple dimensions.
I mean, Josh, you covered it quite nicely. I would say, I mean, also in the Q&A section, if you were tuned in, Mark was also talking about five levels, which allows us to go beyond calling out the four bases. So there are several different directions we can go. I mean, and we really see this as a generational technology at the end of the day, and we wanted to come out when we truly felt that it's there, that it can be a generational technology like many of the other technologies that we have in Roche.
Squeeze in a follow-up, if that's okay?
Yeah.
Okay. Great. I was going to ask, you talked about both RUO and clinical diagnostic use cases. My interpretation, I'm pretty sure I got this right, that initially it's going to be RUO setting, but then you plan to expand down the road. Could you just talk about, I think that's going to have to be separate boxes. It might be more or less the same box, but just sort of with a slight variation to it, but opportunity to expand the platform beyond that. You're talking about mid to high throughput. Could there be a low throughput offer and just sort of what do you see as the potential to go beyond that initial RUO offering?
So I don't think we're in a position to talk about subsequent instruments at this point in time. But I think what we can clearly say is that we do intend to take this technology IVD. And then the second thing I would say is we already see scalability and flexibility of throughput in the platform as it is. So when the gentleman earlier was asking about segments of the market to go for, I mean, to be honest, we see this applicable in a wide variety of throughput settings already in its current instantiation.
One thing I would add is we should think a little differently about this technology than some of the other technologies that are there in the market because it's quite modular. If you were kind of listening to the webinar, it doesn't necessarily mean that every time there has to be a new box. It's quite modular in the sense that we can have step function improvements by looking at some of the key pieces that are there, like the sensing module that sits within the sequencing instrument. There are different ways that we can really continue to push the technology without having to be thinking in a traditional way of launching new boxes the way the sequencing market thinks today.
That's really helpful. Thanks so much.
The next question goes to Doug Schenkel from Wolfe Research.
Good afternoon, everyone, and good to see you again, Birgit and Josh. I guess I have a three-part question, Josh. I think it's for you, but obviously anybody else, feel free to chime in. The top three things that jump out at me that I think need to get better are, one, reducing sample prep requirements for duplex sequencing, which I think are quite high. Again, I want to get your take on that. Second, the need to get the Q-scores up when you run simplex because Q20, for a lot of applications, not all of them, but it's pretty low. And then third, presumably back-end interpretation needs to continue to advance, and I make that assertion just seeing what's been a challenge for other nanopore-based approaches in the past.
It is a lot of data and a lot of reads that you're generating in a short amount of time. That's great, but you got to be able to handle it. Those are the top three things that jump out at me. I'd love to just see if you agree, if there's reasons I'm wrong. If I am wrong, then what is holding you back from launching today? Basically, if these aren't the things you really need to optimize before launch, what are the other things that you're really needing to get better at before you launch as an RUO at some point a year from now? Thank you.
Thanks, Doug, and good to hear from you again. I didn't say it earlier. Nice to see you again, Subbu, as well. Thanks for the question. The three parts you mentioned, sample prep for duplex, Q-scores on simplex, and back-end interpretation, do I agree? I guess I would say I'm not really sure I agree. I'll try to give reasons why. Then, of course, I understand then you're going to say what's holding us back. Sample prep for duplex is something the team is spending a lot of time on and optimizing, but it's pretty straightforward. We haven't disclosed a tremendous amount of detail on that. That's something that I think obviously customers may speak about at the AGBT workshop. We can share more over time. This is not necessarily that complicated to do.
And so I would say when we think about our time and our workflows, I mean, as you saw with the SBX fast example, that was duplex sample prep, and it didn't take a long time. So I'm not sure I entirely agree with that one. On the Q-score for simplex, that's really for customers to decide and not for me to say. But I would encourage us to think about simplex only for simplex-relevant applications. And so if you're doing a counting application, I don't think higher Q-scores are needed. And obviously, if you do want a higher Q-score, the duplex mode is what we would suggest as the primary mode for those high-accuracy applications. So we really are thinking about this as two workflows on a single platform. And I think customers will select what they see as fit for those applications.
So that's my reason why I'm not sure higher Q-scores for Simplex are needed. And then on back-end interpretation, just to clarify, we've already used tools and are already developing our Accelerated Compute Module to deliver the variant calling accuracy that was presented during the webinar. So that is part of our solution. Now, variant calling, as we said on the Accelerated Compute, it's not variant calling on there right now, but that's standard variant calling analysis, right? We produce the same kind of files that the industry is used to. And so this is not necessarily anything novel or new. So I don't think back-end interpretation will be really a challenge either. We have already addressed most of all of this, and we will continue to optimize even further from the starting point we're at today. So lastly, thanks, Doug. You asked what's holding us back.
This, to me, is just a simple function of how Roche thinks about launching a quality product. We are in early access, but we intend to do a lot more work to make this technology as robust as possible. We think the reproducibility is quite good now. But as we are optimizing and preparing ourselves for launch, that's why we don't want to just throw it out at lower volume or without the optimizations that we will do between now and launch. Anything to add, Palani or Matt?
I think you covered it quite nicely. I mean, at the end of the day, at least from my side, I mean, it's just to the very last page. I'll just double down and say that this is part of standard product development because we are always introducing products at scale. So we have to always make sure the reliability, the robustness, those testings. So it goes through a standard verification and validation process that we have to go through, and that's what is the process that we'll continue.
Yeah, totally agree. And I also appreciate Josh saying it's for the customers. They like the counting applications or longer reads, longer reads. I think those are really things that they're going to let the application drive which assay format that they go with. And I think that the power of the high-accuracy duplex, like you said, the workflow was shown to be quite quick. So yeah, very well answered.
Okay. Thank you, guys. Is it possible if I could just sneak in a really quick one? And that was super helpful. I believe the data you presented is in, and again, I might be wrong, and I want you to correct me on anything you disagree with or I'm wrong on. That's part of the point here. But I believe the data you presented is on highly conserved regions from a sequencing standpoint. Josh, you and I have, we've both seen that from other sequencing companies. And then when you actually see the data across all common regions, there can be quite a difference. Is there any concern here as we look at the data you shared today and we kind of think about that?
I'm so glad you asked that question, Doug. I do know what you're alluding to, and let me make this as simple as possible. All the metrics you saw are assessed across the entire known true set of human genome references. What that specifically is, if anyone wants to look it up and check it, is what's known as the Genome in a Bottle high confidence region. I think in the footnotes of our slides, we've even annotated exactly which version of the high confidence region we used. But to make it very clear, it's not a subset of the high confidence regions. It's the entire high confidence region. So that is an apples-to-apples performance characteristic across any of the sequencing technologies.
And those who have followed the evolution of high confidence regions over the years will know that that actually includes known difficult regions, not just from short-read technologies, but also long-read technologies. So it is our best knowledge as to what the true set is for variant calling.
Super helpful. Thank you.
Welcome.
Great. Then the next question goes to Sung Ji Nam from Scotiabank.
Hi. Thank you for taking the questions. Just a few clarification questions. When you talk about an open platform, there are different levels, right, of openness in a platform. So I'm just trying to figure out if you guys might be targeting the most sophisticated sequencing users when you initially launch, or could this be adopted by anyone with any familiarity with any type of sequencing? And then just on the instrument platform, are there any special accommodations necessary in the laboratory for the platform? And also, could you talk about kind of the footprint for the instrument? It looks pretty compact, but just obviously can't tell how big it is. So just curious on that front as well. Thank you.
If you don't mind, I'll go backwards through those questions and start with the last one first. Like with all other specs, we are not releasing any final specs, but it won't be a huge mystery. Anyone who's at AGBT will be able to see, we will have physical versions of both the synthesis instrument and the sequencing instrument in our booth at AGBT, and so people will be able to get a sense of the size there. I don't want to make any qualitative statements that could be misinterpreted to it. I'll let people look at it and see for themselves, but we don't think these instruments are overly burdensome in their size. We think they're nice and compact.
Your question around what type of customers we're going into and the level of open system, your first question there, I would say it's actually really just a function of user. There's nothing about the way we are intending to take this platform to the market that would restrict what the users could do versus other sequencing platforms that they're used to in the industry. That means the most advanced users will be able to not just use the platform, but even invent new methodologies and ways to take advantage of the sequencing technology. New library preps can be put on the sequencer. It's not closed that you can only use it in a certain way. They can also use third-party tools that they already use, be those reagents or software for library prep or bioinformatics. Obviously, many tools can be optimized for different technologies.
So I'm not going to say all of them will work in exactly the same way, but there's nothing stopping anyone from using their favorite library prep kits or their favorite software with this platform. So it really just depends on the user how much into the weeds of optimizing different parameters they want to get. If they want to run things in a plain vanilla way, it will work for that as well. And eventually, as Palani presented, we will also offer our own end-to-end solutions to make it as easy as possible for users.
Also, I mean, I'll just add that you asked the second question about special configurations in the lab. We don't see that. I mean, to the earlier point that Josh was mentioning, the accelerated compute for primary and post-primary analysis that comes with the instrument. And then the secondary analysis is customers can use their own pipeline for variant calling or so on, but can also, of course, use some of the open-source pipelines that we will make available.
Great. Thank you so much. Okay, great. I think we are almost at the end of the hour. And also, we don't have any more questions in the queue. So I would like to thank the presenters very much and the audience for your interest in our sequencing technology. If you have further questions, please reach out to the IR team. We are very happy to assist you. So now I would like to wish everybody a very good day or evening, depending on where you are. And thank you very much for dialing in. Bye-bye.
Bye-bye.
Thanks, everyone.