Hey, everyone. Good afternoon. My name is Tejas Savant. I'm the life science tools and diagnostics analyst here at Morgan Stanley. Before we get started, for important disclosures, please see the Morgan Stanley Research Disclosure website at morganstanley.com/researchdisclosures. If you have any questions, do reach out to your sales rep. It's my pleasure this afternoon to host Oxford Nanopore, and from the company, we have Gordon Sanghera, CEO. I know, Gordon, you want to make some opening remarks and introduce all the shiny toys you've brought with you. So, take it away, and then we'll get into Q&A.
Thanks, Tejas, and good afternoon, everybody. I might be the last person between you and the drink this evening. Hopefully, we can entertain you. I have here some toys indeed. I have some sequencers, actually, Nanopore sequencers. And I'll start with the MinION, which Oxford Nanopore was formed in 2005 and was launched 10 years later. And it is a small, portable sequencer, plugs into a laptop. Business model is consumable sales with a flow cell and a kit. Coupled to that, we can either plug it into a laptop or we can make a completely portable integrated version. So this is basically an NVIDIA compute, an iPhone screen, and a MinION flow cell here. Last year, nine months ago, we launched this. This is a two-channel PromethION sequencer, and it has the flow cell in here.
It's six times the density of electronic channels and therefore the throughput compared to the MinION. And it simply leverages Moore's Law to be able to give you that increased throughput without an increase in cost of goods. These are all electronic sensing devices, not optical, and for each of them, we basically have three customer types. So our smallest group, our genomic entrepreneurs and explorers, spend up to $25,000 per annum, medians, about $6,000. These people are using Nanopore, where we read native DNA, to rewrite the rules of biology. Because when we look at native DNA rather than sequencing by synthesis, we see the full biology, that genome in high definition and full color. Our second group are MinION GridION users, who typically a GridION-- a GridION is five MinIONs in a box.
So they tend to be medium-scale users, 25-250,000, with a median spend around $60,000. Public health laboratories and more mid-range, mid-scale programs, panels and so on. The third group are our super users. They have multiple devices, and they either have a P24, which is a 24-channel version of this, or a P48, which is a 48-channel version of this. They spend greater than 250,000 and typically $600,000. We have six or seven thousand genomic explorers, over 1,000 customers who are in the mid-range, and about 90 multi-device P24, P48 customers. In each case, the commercial model is to sell these as bundled OpEx, so the device is free.
But our cost of sequencing is comparable in cost per gigabase across the segments, whether we're going head-to-head against NextSeq to MiSeq, or at the top end, NovaSeq, but without that upfront CapEx. We can read native DNA of any fragment length, from very short, which is really interesting in liquid biopsy space, to long, tens of thousands of bases, to ultra-long, hundreds and thousands of bases. So it's a technology that can do long, medium, and ultra-long reads with native DNA. These are all highly differentiated. One final thing, this is real-time live streaming, and it's not batch-based, so you don't have to maximally load this machine. You can put on one sample on a P48, or you can run 48 samples. Your volume-based discounting will give you the same cost per genome. With that, I'll hand over to you for questions.
Sure.
Tejas.
So, Gordon, it's been quite the journey for Oxford Nanopore over the past, you know, decade or so, really. And, I've seen it sort of evolve from being viewed as, you know, fun, hip, cool, but ultimately niche and slightly unpredictable platform, to one which has really sort of matured now and, and, and come into its own, right? And, and part of that has been, you know, improved accuracy, and we'll get to that in a minute. But it'll be great to just get a high-level overview of, you know, what's gone well for you over the past sort of few years, and, and where perhaps you could have done things a little bit differently in retrospect.
Sure. So I think, with such a different offering to sequencing by synthesis, it's not just a small step. It is really a big disruptive step. And we, we chose to launch the MinION, this $1,000 starter pack, to get technologists, customers, to prove out two things: the value proposition, the distributed benchtop students running simple sequencing devices was something that was of interest commercially, and secondly, to really drive technology validation. And, and that has led to over 8,000 publications extolling the virtues of native long-
Mm-hmm.
DNA. And so in the last 2-3 years, there's been an exponential growth. I mean, the pandemic resulted in public health laboratories having to sequence COVID variants.
Mm-hmm.
The GridION, five MinIONs in the box, really drove adoption of nanopore sequencing in low- and middle-income countries. We sequenced over two million COVID genomes. At the start of the pandemic, we shipped 200 MinIONs-
Mm-hmm
-to Wuhan, and today, the Chinese CDC have this as their surveillance monitoring for next pandemic. So things have gone okay in that period. What was challenging was scaling-
Mm-hmm
And getting that reproducibility and always driving accuracy to reach Illumina standards, the gold standard as it is for now.
Mm-hmm.
So that we have managed to achieve. March last year, we launched our Q20+ chemistry. The reliability and the consistency across all of our platforms is here now.
Mm-hmm.
Yeah, we are growing up and becoming a mature company.
Got it. Yeah, so that, that's actually a perfect segue into my next question. Can you walk us through what you've changed to overcome, you know, some of the earlier issues you saw on systematic bias, et cetera? You've got the new chemistry, obviously, but there's also other stuff that you've done to address some of those challenges. Just walk us through that.
So the -- let's deal with the systematic bias.
Mm-hmm.
There is with Nanopore sequencing homopolymers, and today that is now down to a very small range of homopolymers, greater than 10 bases, 10-12 bases.
Mm-hmm. Mm-hmm.
We still face a little bit of a challenge, but in the context, for example, of human genomes, they're in non-coding regions. So we have now pretty much removed systematic bias driven by homopolymers. In bacterial genomes, there are very few genomes with greater than 10 homopolymers.
Mm-hmm.
So it's kind of done, but we have some polishing steps that we will introduce that will completely remove that. So we are substantially equivalent on accuracy with SBS and variant calling. It's been driven by improvements to the nanopore, because that increases the signal-to-noise. Quieter electronic motherboards, so the electronic noise is lower, but most significantly, machine learning AI.
Mm.
The exponential rate at which we're seeing that technology progress, we can just leverage on the back of that and take those improvements and port them over. That has been the significant driver in getting our accuracy to what we call the Q20+ chemistry, which we launched last March.
Got it. And what's been the early user feedback on the new Duplex Chemistry, and what is the accuracy that customers have been able to consistently achieve in their own labs?
Before we talk about Duplex Chemistry, I'd just like to talk about the Q20+-
Sure
-Simplex.
Sure.
So today, we did this rollout in a very controlled and limited manner, probably the most mature rollout of a product. Very similar to my old company, Abbott, very careful about how we put it out there. It's widely adopted now. In the context of a human genome, you can get Q47 accuracy. On the bacterial genomes, it's greater than Q55. So customers are replicating on Simplex that same accuracy. Now, Duplex, for those of you who don't know, is where the first strand is followed into the pool by the second strand, so it's pairing. So we can take both in majority call, and that gives you Q30 single-pass accuracy. Now, that chemistry is really important for de novo sequencing-
Mm-hmm
... and for creating telomere-to-telomere reference genomes. And while a year ago it took Nanopore, Illumina, Bionano Genomics, and PacBio to sequence a full telomere-to-telomere genome, a year later, and that cost about $25,000. It was done by, led by Karen Miga at University of California, Santa Cruz, and Nature published it as a Method of the Year.
Mm-hmm.
Long reads. Today, you can do a telomere-to-telomere with Nanopore only, using long and ultra-long, for a tenth of that price.
Mm-hmm.
That gives you the most comprehensive human genome reference that's out there today.
Got it. On the last earnings call, you talked about, you know, the importance of, you know, the A Series compute upgrade that you're rolling out on the PromethION. I think you talked about a 4.5 increase in base calling speed. But what does that translate into for the average user in terms of just, you know, cheaper, fully loaded cost per genome?
Sure. So the A series upgrade enables us, with the launch of, our methylation base calling software, Dorado, which is integrated into our operating system, MinKNOW, to be able to do real-time methylation... as well as whole genomes. And that's really important because customers really want to see that 95%-96% of the methylome that we can map. And this is not an additional run step of several hundred dollars to do bisulfite sequencing. So that really brings all that together, and it also, that compute power, allows us to go to more complex machine learning models to deconvolute-
Mm-hmm.
and push the accuracy envelope as well. So those are both significant drivers in commercial adoption.
Got it. Can you talk a little bit about where you are in upgrading your customers to the R10 series, and what are the implications of that for instrument margins? And how does the product sort of stack up versus, you know, reagent on board from Illumina or some of the other solutions out there?
Because it's very apples-to-oranges comparison, but because our data streams live in real time-
Mm.
the upgrade to the A100 series and that much, much bigger compute capacity for us to have more complex machine learning
Mm.
signal processing algorithms, it's a no-brainer for us to take a short-term margin hit.
Mm-hmm.
-to upgrade our customers because we are very confident that it will grow, consumable adoption.
Hmm.
As we look, particularly on the PromethION fleet, at Q20+ chemistry, we're now at 90% adoption of the Q20+ chemistry. On MinION, it's about 50%, but that's because the R9 series chemistry accuracy is sufficient. So accuracy is not a binary.
Hmm.
There's a continuum.
Yeah.
And also, there's a lot of legacy projects that people are finishing. In the next six months, we will completely switch out the 9 series to the 10 series. The short-term hit on margins will pay in spades in the medium term with consumable pull-through. And on that point, 75% of our revenues-
Mm-hmm.
-are consumable.
Mm-hmm. Got it. Two other things I want to hit on before we move to, you know, the product portfolio side of things. Short fragment mode, what is, you know, uptake been like on that front? And in terms of adaptive sampling, what types of applications do you think this could be suitable for? And can you comment on some of the traction you're seeing for that to date?
Short fragment mode.
Yeah.
It's really exciting. We've always been able to read short strands of DNA, but our focus was to produce a highly differentiated unmet need in the long read space. But we have turned our attention to short fragment mode, and we launched the product about 18 months ago. The first thing that's really interesting and significant, because we're looking at native DNA, and we can read any read length-
Mm.
So whatever's in the sample, because we're not cutting it down to 200 base pairs, circulating tumor DNA can be as long as 3,000 bases. So that's a bit of a shock for everybody. But more importantly, and this is some great work done by Hanlee Ji at Stanford, he looked at colorectal cancer patients and controls. Those short fragments on the cancer patients had 7% methylation. The controls had 2%.
Mm-hmm.
Then he had some longitudinal samples where he showed treatment efficacy, and the 7% methylation dropped to 2%. But then, as the treatment... the patient became resistant to treatment, it goes back up to 7%. So I think that's really exciting. We also are the only company that can read RNA directly.
Mm-hmm.
So not cDNA. And so there are... cDNA leaves certain parts of RNA unread, dark RNA.
Mm-hmm.
And, there's a publication from the University of California, Santa Cruz, where they're looking at this dark RNA as a predictor for cancer in liquid biopsy. So I think that space is really going to be interesting, but we're in the foothills of that journey.
Hmm.
It's the really discovery phase that we're in. I do think it will be exciting to be looking at native DNA and modifications in liquid biopsy applications.
Got it. Are you—you know, following the launch of the Q20 chemistry and the R10 pore, are you starting to see that inflection in the install base for the PromethION and the GridION?
So, it's interesting. I mentioned that the, particularly in the S3 customer base, so PromethION P24-
Mm.
P48, in that grouping, they are rapidly adopting.
Hmm.
Not surprising. They're the most sophisticated, and they're doing the highest—
Mm.
-hardest biology. And so, yeah, and, and we do think that we posted for half one 47% underlying growth.
Mm.
So stripping out our Emirati Genome Program and COVID.
Yeah.
That, we think, is, you know, a big chunk of that is driven by PromethION sales, PromethION flow cells, and driven by that Q20+ chemistry, and the upgrades to the platform as well.
Got it. What's the latest on VolTRAX, you know, the automated library prep solution? What % of your GridION and PromethION users are leveraging that for library prep?
... So VolTRAX has been a interesting project for us. It was a collaboration with Sharp Life Sciences. About 18 months ago, we acquired the asset, and we pretty much put on hold what we were doing with Sharp. We are completely revamping the offering on VolTRAX.
Mm-hmm.
We have a few legacy customers we can support-
Mm-hmm.
-but it really is a project that is back in development, and, you'll hear more about how we are reconfiguring it and refactoring it.
Got it.
In development, not in production.
Got it. And do you view that as an important solution to offer to your customers that could really accelerate adoption?
So when we think about crossing the chasm to the applied markets, I think automated sample prep solutions will be central. Whether that's, and we announced a collaboration with Tecan-
Mm-hmm
-on our automated liquid handling.
Mm-hmm.
Sort of classical robotics or very miniaturized for decentralized point-of-care setting, which is where VolTRAX will sit. We do think they will be central in moving sequencing from centralized-
Yeah
specialist, multimillion-dollar facilities into broader application in the decentralized settings.
Got it. You talked about, you know, 500 customers with 20% new accounts for the P2. Where are you seeing the strongest uptake? And as you think about your portfolio, where does this fit into, you know, the much larger PromethIONs?
I don't know. This is, this is an interesting beast. It's, it's kind of if you use a computer vernacular, it's, it's a supercomputer in the hands of really creative people. So it'll be interesting to see where it lands. In terms of the 500 placements, the first nine months, it's across the S1, S2, and S3 customer base, and
Was that surprising to you, Gordon?
I think, well, when we asked, "What would you do with a 2-channel?" We were surprised that the S1 customers were interested. So I think, you know, the MinION comes in at $1,000. This is $10,000 dollar pack. I don't know what their ceiling is-
Right
... for, you know, entering the game, and maybe it's 20-30,000, so this will drive sales over there. But of the 500, it's too early to see exactly where they are and what they're doing. But what is significant and interesting, if you run this at full tilt, it can do $170,000 in consumable pull-through per annum.
Hmm.
We've actually got a few outliers, I don't want to mislead anybody here, who are actually running these at full tilt, based on utilization. I think it'll come in somewhere more like the S2 customer grouping, around a third of that-
Mm-hmm.
$60,000. But you know that. I think the thing that happens with this is you, you can really start to do some chunky and complex biology, which allows you to make small data sets that allows you to get bigger grants. That's how we've seen MinION progress, and this is just a super, super version of that. And I think that can then really lead to an accelerator of P24, P48, but not necessarily, because you can daisy chain a handful of these-
Mm-hmm
and run off those as well.
Got it. So just to make it more real, what would, what would this translate into in terms of throughput or samples and for, let's say, a targeted sequencing panel or something like that?
Let's be human.
Yeah.
Because that's the answer I rehearsed. So on a P24. Let's do P48, because I know the answer to that one. P48 is 10,000 per annum.
Yep.
So you can divide that by 24, and you get down to that. But we have a roadmap that allows us to go from one genome per flow cell. It's actually one and a half you can get at the moment, and you can do one and a half, but 1-2, which means you get to 20,000 genomes per annum on a PromethION 48, 10,000 on a P24, and then a few on there. And that roadmap then leads us to three genomes.
Mm.
We have a pipeline. We are increasing that scalability and throughput at a dramatic rate, and we will continue to do so. That's where the focus lies now that we have the Q20+ Simplex chemistry and the Q30+ for de novo and reference genomes, but the bulk of revenue will come from that Simplex, and that's highly competitive in cost of genome as well.
Got it. In the interest of time, I'm going to ask you sort of to pick, pick, you know, a favorite among your children. So direct RNA sequencing, you know, the ultra-high throughput PromethION. I think you're doing something with, with ASIC as well on, on the lower cost, low power devices. And, and protein sequencing is something you've, you've talked about, I mean, not, not so shared so any data around yet, but that's, that's in the hopper as well. So which among these do you see as the most, near term and, and needle moving?
Well, I love them all. They're all my children. I'll just knock them off in reverse order. Protein sequencing is coming. We're really excited, and we have translocation, and we can measure some of the peptides, not all yet.
Mm-hmm.
But the chassis, so I think it'll be in this beast, the chassis is already built. It really is just changing the operating system, which is the nanopore. Now that we've solved the translocation, it's really the reader head that gets us there. The low-cost ASIC is really exciting. I come from a point-of-care blood glucose background. We help to decentralize and democratize blood glucose analysis for type 1 diabetics at home. I think the low-cost, low-power ASIC allows us to get to single use.
Mm-hmm.
I think that's critically important in industrial applications, whether that's real-time biologics manufacture or some of the clinical applications we're excited about. What, what were the other ones on your list?
Oh, they were basically the direct RNA sequencing. That was-
Direct RNA will be a game changer. We are at sort of translocating at 200 bases a second. That's five times what we were doing a year ago, and we think that methylation paper on that liquid biopsy is a great example of how that field will emerge. Yeah, ultra-high throughput. Ultra-high throughput, T to T.
Mm-hmm.
Every week there's a new reference genome. We think that will become the new gold standard.
Mm-hmm.
And that's what everybody will be judged against. And this SBS, 92% of the genome as a gold reference, will become a thing of the past. So that's also important. But they're all important. They're all massive drivers in pushing growth through innovation, which is what we're all about.
On the PromethION today, Gordon, what's the price point like for genome at 30x coverage? And, is it this ultra-high throughput offering that gets you to, you know, I don't know, sub-$400 or perhaps even lower?
One genome today is $690.
Mm-hmm.
We're a few months away from putting two genomes on each flow cell.
Mm-hmm.
That gets you three, four, five.
Okay.
This is 30x simplex. That's all you need. And if I could just comment on, just correct some things that may or may not have been said today about storage costs.
Mm-hmm, mm-hmm.
There is. You can have raw data, which is the raw signal as we generate it, before it is then converted through machine learning AI into sequence data. So, industry standard sequence data files are. Ours are equivalent. Storage costs are equivalent. The raw data is a unique feature to Nanopore, and the way to think about that is, the data is not dead on arrival. What I mean by that is, for the highly sophisticated users, they can retain that data.
Hmm.
The storage cost is not $600, it's $15 on AWS. Then they can go back and re-basecall it with modifications or upgrades or new insights if they wish to. That is a feature for the raw data.
Mm-hmm.
I just wanted to clarify that. As we go from $3, $4, $5 on two genomes to three genomes per flow cell, we get closer to $2-$2.50. And we think a fully loaded, full methylation, hot long and ultra-long read structural variation, copy number variation, and the SNPs and variants you get with short read, that sort of $200-$300 is a sweet spot for fully loaded genomes that are the gold standard, the new gold standard, and that's where we, we are setting out our stall.
Got it. I'll ask you a sort of three parter on the competitive landscape, right? So you've got PacBio, I mean, they started shipping the Revio earlier this year. How often do you run into them on RFPs, like particularly for your S3 customer base? Or is it still sort of largely non-overlapping? And then, you know, to the extent that you do, when do you win and when do you lose out? You know, what are the sort of customer criteria that sort of work in your favor, and when do they work against you?
With the RFP overlap, I don't think we see that-
Mm-hmm
... Because of the OpEx model.
Mm-hmm.
We're not feeling any pressure. But let me give you one or two examples of where we are. So, we are exclusively in a 4,000-patient trial with NIH on Alzheimer's.
Mm-hmm.
They need long reads. It's well understood that long reads and structural variation go hand in glove. They've chosen us because of throughput and scalability, and they recently published their first data set showing that they were getting the same structural variation comparable to HiFi reads, but in a highly scalable and more cost-effective manner. On the other side of the pond, in the U.K., Genomics England have what they call Cancer 2.0.
Mm-hmm.
Which is to introduce sequencing, human genome sequencing into hospitals, into the NHS, through their specialist labs called the Genomic Laboratory Hubs, the GLHs.
Mm-hmm.
They chose Nanopore over Illumina because the long reads and the native DNA give you the methylation structural variation, which you need for cancer. That is exclusively several thousand patients on Nanopore, and there are multiple opportunities. We focus on ourselves, we focus on the low-hanging fruit, where we think we have a unique value proposition because of the native, all the long read, all the short fragment native, where we believe that we can provide value, and that's how we focus on getting those opportunities. Q20+ is a big accelerator of those-
Mm-hmm
... Opening those doors and having those conversations.
Got it. Competition for the marginal research dollar, you know, even sort of outside of long reads, there is a focus on, you know, spatial genomics or perhaps proteomics, you know, multimodal approaches. How do you think about, you know, the where in that sort of, you know, priority list does long-read sequencing or long-read sequencing via Nanopore lie for, you know, your customers?
We are heading to a multi-omics world. Our view is, so we've got SNPs, SMBs.
Mm-hmm.
And then we've got structural variation, and we've got copy number variation. Those are multi-omic. On top of that, because we can read full RNA transcripts-
Mm-hmm.
And our sample prep is not two days, it's an hour and a half, and you get that full transcript. We have a marketing partnership with 10x Genomics. People are, and it's highly scalable and highly competitive. You get full transcripts at the same number of reads as you get short transcripts. That's really exciting for us. And then the next thing we plan to do, and we are making good progress, but we'll talk about it when we're ready, not before, is proteomics. So we believe the Nanopore platform will offer that multi-omic play all the way across with no platform shift.
Hmm.
All on the same sensing platform and measurement systems. So yeah, we're really excited about that whole menu and the offering that we will bring to the table.
Got it. Quick clarification. Direct RNA, that's, that's supposed to go live by year-end. Is that still the plan?
No, that is direct RNA is out there.
Oh, okay.
It's available.
Okay.
I think where we are on that is it's right at the beginning. People are really-
Got it
... trying to understand. There are many, many modifications on, on RNA.
Right.
People are starting to look at direct RNA and get their arms around that. The one exciting thing about direct RNA is we could play a significant part in vaccine manufacturing-
Hmm
where modified RNA is inserted to stabilize the vaccine.
Got it. Last question, I'd be remiss if I let you escape without commenting on China. It's been something that's come up a lot over the last couple of days, in terms of just, you know, the weakness there on the pharma side, also the anti-corruption crackdown. What is your China exposure, and are you running into any of those headwinds in July and August? I know the first half was really good, right? So I think you called out 20% growth in China.
Yeah. So China, China is about 9% of our total revenues, global revenues. In the first half, we saw if you strip out COVID-
Mm-hmm
We had a huge headwind. They were doing a lot of COVID sequencing, and they just stopped in January, as we all know. So our overall growth was greater than 25%, and we're very pleased with that, but we remain cautious, cautiously optimistic about China.
Got it. And nothing's happened since that's changed your mind about business conditions in China?
Not yet.
Okay.
It's rapidly evolving, you know, so we are vigilant about-
Right
what may happen.
Fair enough.
But at the moment, underlying, we were very pleased to be greater than 25.
No signs of economic contagion to Europe. I mean, particularly, you know, markets like Germany are export-oriented with the sort of-
No, we hired some senior leadership from Illumina to run Europe for us.
Mm-hmm.
Actually, we won a long DNA sequencing program in Germany recently. It's hard to get at these programs, and when you do, they can be game-changing. No, not at all.
Great. That's a great place to leave it at. So appreciate the time, Gordon. Thank you.
Of course.