SOPHiA GENETICS SA (SOPH) Earnings Call Transcript & Summary

Good afternoon, everybody. My name is Luke Sergott. I cover life sciences tools and diagnostics at Barclays. It's my pleasure to have the man, the myth, the Muken here with me. Ross Muken, CFO of SOPHiA GENETICS; and Kellen Sanger, running Strategy and IR for the business. So the evolution of the business and the platform. Why don't you just kind of give us a sense of the 101 of the SOPHiA DDM platform. I mean, it was -- you started off as like a genomics data analysis package and really kind of how that's evolved across the different demand and customer applications that you guys have built out?

Sure. So a number of years ago, probably over a decade ago, we saw the shift of sequencing to the clinic, right? And with that, we also saw that there were going to be quite a lot of vendors that played across the continuum of that production process of precision medicine data. And what was obvious to us was all of that diversity of instrumentation and all of the ways the data was going to be produced was going to create a lot of noise and it would be quite hard to operate a lot of these sequencers in a decentralized format. And so we thought that a platform, a software platform that would allow for the production of that data in a harmonized way would be something that would be well needed versus all of the different institutions developing their own code and ultimately working on this N of 1. And so we built the DDM platform for that purpose, to allow laboratories and hospitals and the like to be able to produce data anyway they want, any mix of instruments, reagents, automation, et cetera. We'll correct for it on the back end with algorithms. And no matter where you are in the world, you can get pristine highly accurate data. And then it became okay, if you can do that and connect all of these institutions in 1 common network, you should be able to also advance the science by doing knowledge sharing and collective intelligence, right, across all of these different institutions producing data. So it became very much a feedback loop that the more data you shared and the more data that you produced, the better the community insights were and ultimately, the more you can involve advanced care, both in oncology and rare disease. Now the evolution of that became not only were we able to support single gene or 5 gene or 10 gene or 25 gene testing in the early days, then we started to see panels and more significant gene content, 50 gene panels, 100 gene panels. Now we're seeing CGP 500 gene panels where we're seeing exomes in the thousands, right, and eventually now whole genome. And so we provide the analytical support for any type of application no matter sort of the area of interest for rare disease and oncology on the same platform, right? We also have different types of universal library preps that we work with our partners on. And there, again, across all of the different applications you can use the same type of prep. So very efficient, very low cost and ultimately, again, best-in-class accuracy. Now the next sort of iteration of the model became, as we saw the N of the data grow to such a size level, you get what's called network effects, right? So the data you have, you'd start to see things in that data where you can transform it into a new product, essentially that you can then use in another arena. And so we saw the ability to take that data and bring it to pharma, right? And so now pharma is another sort of entity in the network where they're producing data or asking questions. And again, you're getting with that the typical kind of flywheel effect that the more questions they ask and the more data they contribute, the more the network produces and the more the network pushes back to them and vice versa, right? And so that's causing growth. And then the last piece of it and more recent was sort of our journey into multimodal. So then it became, okay, it's really interesting to look at sort of the biomarker level data, but what if you can do that plus the radiology data plus the pathology data, plus the phenotypic data that exists in the EMR and look at that in a sort of multivariant way and be able to draw new conclusions, right? And that's sort of where we've been pushing with the data ingestion and so more patients, more applications, more data content. And so for that, we think the world is in the very early stages, right, of being able to produce and harness all this data. I mean a lot of the investment has been made in terms of the CapEx. If you go into a hospital, they have multiple sequencers now. They have lots of MRI or CT machines. They have tons of anatomical pathology microscopes. But none of that data gets kind of aggregated anywhere or shared, right? And certainly not insights drawn off of it. And so we see in the future a very different view for the patient, for the oncologist or for the clinician. And again, all of that feeding to each other to create this collective intelligence to ultimately advance the care of disease and improve patients' lives and allow people to live, right? And that's kind of the end of it.

And just talk about -- I mean, all these disparate forms of data from the analytical pathology where you're looking at slides versus multi-panel sequencing. It's not all standardized and able to be analyzed. So talk about the upfront, the processing, the investment that you guys had to make to actually standardize that data so that it could be just a single page readout to the clinician.

Yes. This is probably the piece I think in our world, right? You hear a lot about AI now, and it's hard I think for many people to understand sort of what's unique, what's differentiated, what's the scale advantage? What's not? But the reality is you think about how businesses like Google or Facebook, others kind of grew up. It's always the same thesis. You have to have -- you have to capture an N of users, that N of users or folks utilizing the network have to have a certain level of diversity. And then from there, your algorithms get more powerful. And then as you keep challenging those algorithms, they become better than anything else. And then you basically get to a point where your lowest sort of marginal cost per output and your lowest cost to sell it beats everybody else, right? And then you win the whole network. And so if you think about that in our case, right, we've been getting to an N of patients now where we produce more genomic data than anyone else in the world. And then from there, you can then do other things, right, within you win the right to be able to say, okay, my algorithms are super powerful here. Will you let us test those algorithms as well in radiology or pathology, et cetera? And so if you look at what we've done with DEEP-Lung and other initiatives, these were sort of the early stages of us showing that our algorithms are now so powerful, right? We can apply them to different data sets, right, or data types. And again, there's obviously evolution of the algorithms in each of the different settings, but the principles and kind of the backbone of the architecture supports it. And so the $400 million, which is a lot of money that we put into this in the 10 years, really now allows us to benefit from those network effects as we scale the architecture. And there's an argument that as we grow bigger, we actually become, right, more differentiated and more unique and theoretically can expand the growth because you get to a certain size where there's no single lab or institution in the world, right, that could not [ trust ] on accuracy or cost or utilization.

And then so on the penetration, the market penetration, as you guys see it, how many -- what -- how many labs can you -- as you look at the landscape, like could be users and where are we on that trajectory?

So you know obviously the next-gen sequencing landscape incredibly well. So if you look at Illumina's numbers, right, and you look at the low- to mid-throughput market, you're talking about thousands of laboratories, right, around the world. That, I would say, is our sort of target customer. Typically, they're in academic settings. We have also quite a lot of reference in central laboratories, specialty labs. Increasingly, obviously, this pharma and other, but the sheer number of places where sequencers are present continues to grow every year. And not only that, the power of these sequencers expand. So what you're willing to do on a sequencer or the type of work you can do in-house has changed materially, even in the last 2 years with the introduction of the X and some of the other competitive boxes. And so we think all of that dynamic is quite favorable for us because the world we foresee has a lot of diversity of instrumentation in consumables, which is good for a player that deals with diversity well. And then we can also apply that to many of the other modalities.

Let's talk about with that comment on the X if people want to keep doing it. Are you seeing more mid-throughput customers move upstream and essentially further centralization of that sequencing model? And then, obviously, they're stealing those volumes from the mid-throughput but also their old [ NS6 ] volumes going to the X. Like give us a sense of what you're seeing from that shift in the market?

And it's obviously not just them, right? We see Element, we see Complete. We see PacBio, increasingly Oxford. There's quite a number of players in the market that have very differentiated offerings. And again, Illumina remains the predominant player, but it's a different mix than it was in the past. Again, they have fantastic technology, they all do. We're agnostic. But ultimately, we're seeing different use cases for different applications. In terms of some of the higher throughput boxes, I think what's ultimately happened is the breadth of centers that have adopted these are probably broader than you would have guessed, right? So we see an entirely unique spectrum and all over the world, frankly, of different people that have upgraded their sequencing capacity. Now with that, it doesn't mean all of them are obviously running full volume 24/7 on these boxes, some are only able to run one lane at this point. Now our guess is, over time between the fact that you're seeing people move, let's say, from exome to whole genome or they're moving from a myeloid panel or solid tumor panel to a CGP, that alone will drive some level of absorption of the capacity. And then obviously, as well, just the sheer growth in number of patients being tested. I mean, in a lot of areas, there's still quite a lot of folks, right, that are not being tested on next gen. And so there's still quite a ways to go there. For us, at the end of the day, we're fortunate. We just want more data output, right? And so that's happening. Now the challenge for some, right, is a mix change, right, from a COGS perspective in terms of the cost of producing precision medicine data. I would say of $100, it used to be maybe analytics, would get 5% of it. Now that the cost of sequencing has come down and even the cost of the boxes have come down, the software piece because of the data proliferation is actually becoming more and more provident. And so we would expect that of that $100 to take a much bigger slice of the pie over the next 5 or 10 years.

And when you think about the shift here, I mean, you're talking about a lot of your presence is on the academic or translational. Talk about how you think that this is ultimately going to shift into the actual clinic and like? How you guys are ultimately enabling that?

Yes. So I guess we have to be careful. I mean, we do -- clinical means different things in different parts of the world, right? Like RUO here is quite different than RUO in Europe, for example. And so in many parts of the world, we're used in the clinical setting, right, even though the products may not be labeled. We do have products in Europe moving towards IVDR. We'll obviously have to think about sort of the new legislation here in the U.S. around LDTs. And what we do from an FDA standpoint, similarly, as we enter Japan and other countries. But I would say for the most part, majority of diagnostics done in the world in the clinical setting are RUO at this point or LDT. I think it will probably be some mix like that in the future. What that looks like, we'll have to see. But for us, we can support any sort of setting that exists. Again, we designed the platform under a QMS system and with what's called design controls to be able to have it ready for any regulatory environment that evolves. And I would say Europe has probably been here at the forefront, frankly, of regulating products like ourselves, but there's still a lot of, I would say, evolution that has to happen in many of these different markets.

Yes. And so the MSK, when 360 went off. Now it's the -- was it the ACCESS launch. So dig into that. I mean you guys, on the fourth quarter, you guys had a lot of announcements. So just kind of walking through and how it's like driving this transition into the deeper and deeper clinical market. So what is the DDM plus ACCESS, like ultimately, what is that? And then what is that answering...

[ What is that box ] around liquid biopsy?

Yes.

Sure. So the MSK partnership has been really exciting for us. Sometime last year, we started talking to MSK about decentralizing and deploying a few of their tests globally. And so they actually selected us to be one of their partners to take the test that they're running in-house for liquid biopsy, which is MSK-ACCESS as well as their solid tumor test, CGP, which is called MSK-IMPACT. So they selected us because of our cloud-based platform, our ability to replicate the analytical performance that they have with in-house across the world. And so this has been a huge growth driver for us. An area where we see a lot of excitement, specifically liquid biopsy and MSK-ACCESS tests. So we launched a privilege access program or early access program with a few customers late last year in December, and we'll have that full commercial launch in April. And we're already seeing a lot of demand. We've announced customers such as BioReference, Tennessee Oncology, Dasa who are already adopting the solution. They -- first of all, when you bring the MSK name, especially in the U.S., but also globally, this is really exciting. You -- the customer usually perks up. They like the solution. So it's been relatively easy or exciting for them as they're looking to add new applications, but there's also a really interesting biopharma angle. So as we're working with MSK, they've introduced us to many different biopharma companies or bring some more to the table there. And we specifically announced the partnership with AstraZeneca, where AstraZeneca will actually be subsidizing or sponsoring the deployment of MSK-ACCESS globally across new customers. So this is exciting for us, obviously, as we deploy MSK-ACCESS to new areas, but also for biopharma as they have more patients who are being diagnosed and tested for liquid biopsy and eventually eligible for the drug, but also the data coming off of that, that those tests is exciting for them as well.

All right. So like take a step back and just think about the overall workflow. I mean you guys -- your software as a service, essentially, your software that sits on the sequencer in this democratized environment of sequencing. So how does the sample throughput -- like walk us through the logistics of the distribution model that you guys ultimately offer because you're offering a diagnostic test. So you still have to do the isolation, but like the lab that -- and DOS is going to do the MSK ACCESS instead of having the samples sent to MSK, you just do it locally? So like how do you think about changes from the logistics that you guys have?

Sure. So I think particularly here in the U.S., folks are much more I would say, used to the central lab model where the sample is sent out, it's processed in the lab and then a PDF comes back, right? So for us, it's quite different. So in our model, let's use, again, one of our customers here in the U.S. as an example. So if I'm BioReference, right, I already have a lot of the logistics that exist in terms of being able to do sample collection. And so I would like to now offer liquid biopsy as a capability with a well-validated test. And so MSK-ACCESS being probably the second most used liquid biopsy test in-market and certainly one that clinically has proven given its capabilities is really robust, is one that would be quite attractive. So if I'm them, what I can do is I go and I contract with SOPHiA, right? We can then go to their laboratory, see their setup, right, understand sort of, all right, what automation are you doing? What sequencers do you have, right? What chemistry have you used in the past? And then basically take our kitted version, right, of the MSK-ACCESS that exists, which is our bundle. And then be able to work with them on their web lab preparation steps after they've done the sort of sample prep, early part like the isolation and other pieces, to be able to use our sort of probe content on their sequencer, right, on the flow cell to go run the sample, right, and get the result. And so where our piece comes in is once the sort of sequencer, in their case, it's Illumina, sort of generates a FASTQ file, that goes immediately into the cloud, right? It is then what's done -- secondary analysis happen, which is kind of the variant calling, then there's tertiary analysis, which is the reporting piece. That then ultimately leads to the end result, which will be sent to a clinician, right, after it's signed off by a board-certified lab director. And so we're doing the piece from the back of the sequencer all the way through to the answer, right? And we do that better than anyone at a lower cost than anyone. And that's where, again, the complexity is happening over time. Given all of the different drugs available in market, given all of what we're learning about different genetic content, given the complexity, right, of what we have in terms of many of these diseases. And so that piece in greater scale is what we do and in their -- we're paid essentially per patient that gets uploaded off of the back end of the sequencer.

And then to compare that to other peer models like Tempus and others that are running this way. What is the scale advantage that you have that they might not have or just a differentiation there?

Well, one, you can do all of the analysis close to the patient, right? So let me give you an example. I'm Tata, India. I before, used to send my samples to foundation, right? And then there would be quite a long wait. I mean just think about the logistics of getting samples from India to foundations lab and back again. Again, it's a wonderful test. But here now, I can be Tata, I have my own capabilities. I can use the DDM platform to basically replicate some of those tests, right, that I want to run hereditary or solid tumor. I can do so in a highly accurate manner in a few days, right? So now I'm getting an answer for that patient quite quickly, pristine results. And I can do it with whatever equipment and other elements I have around my laboratory, right? So now I'm able to lower the cost, I'm able to quicken the turnaround time, I'm able to basically do multiple different types of tests on 1 platform with 1 workflow of prepping chemistry, right? All within the confines of my institution. And so it's -- and then what's fascinating is so now I'm pharma, right? And so in the last week, we were -- I was talking to prospects in Australia, one of the largest cancer centers there, in Brazil, in Nigeria, in France. You can imagine all of these centers now producing their own data close to the patient, where they get to keep that data for themselves. They're not just getting that PDF back. They have the raw genetic data, right? Now, they can take that, mix that with their other data and actually get conclusions on their patient if they so choose over time, right? They own that data. And so from that standpoint, it really enables them to be their own sort of precision medicine lighthouse and a lot of the intellectual capital stays within the institution, which they can eventually do other research or clinical trial work, et cetera. And so -- and the data is now comparable because it's on a common platform across all of those different labs no matter where you are in the world. And the insights and learnings we get from all of those laboratories are also shared across the community. So again, it's a very different thought process versus, again, the send-out where the lab is doing the sequencing. Sending back the PDF, right, and stapling it to EMR and then doing the revenue cycle to collect the reimbursement. So even in the U.S. in many cases, given how reimbursement is firmed up in a number of different NGS categories. Now this could actually be a profit generator for the hospital.

And then so when -- if I think about it from the LDT perspective and a lot of the liquid biopsy players, like why wouldn't -- what prevents them from coming on board and essentially partnering with you on the rest of them?

So this makes a ton of sense, right? So this is something we've seen post the MSK announcement, mainly because -- and we seek to do it with other of our peers. I mean we have another product in MRD, AML that we're currently developing with one of our very large super brand name lab customers in the U.S. that's in the Midwest, right? And here's, again, very exciting in terms of what we're able to do with them. But again, it's open innovation, right? We're doing it in combination with thought leaders in the space. We can also do this with other companies, right? There's a lot of technology being built, right, for the decentralized world, whether there are reagent packages or different other configurations. It doesn't make sense for everyone to build their own sales force and build their own analytics capabilities when we have it already deployed in many labs in the world. So you can see a model where someone takes their solution, puts that on our platform, we algorithmically support it, can push that out to our customers. And now you've got distribution already in all of those laboratories that will adopt tests like this. And so for us, this is another angle. And what we're really trying to do longer term is just enable that production of data at scale, right, in a common manner. Because if I'm the lab director, if I'm the hospital, I don't want to run 5 different software applications for 5 different solutions, right, all with mixed kind of variability and have to change my prep every time that I'm doing a new run, it's very hard. There's not a lot of labor in these labs, right? And even with automation, you can't keep up with the volumes as they're happening and the complexity.

And so when you're thinking about almost being plugged into like an epic or the back-end analysis piece of that of a massive health care system software piece?

Yes. Because for us, again, we are a B2B company. And so with our positioning as kind of a network player, you want to plug into everything, right? So we'll go into an institution, we'll plug into your LIMS. We'll plug into your EMR, we'll plug into whatever configuration of equipment you have, whatever reagents, whatever other software you might be using in the laboratory, it's very flexible, right? And the point of that is because what we want to be able to do on our platform is enable everyone to sort of be more efficient, right, and be able to produce data at scale. And so to do that, you really have to sort of weave stuff together to create a solution and a workflow that's better than what exists. And I would say it's time savings, cost savings and the magnitude, depending on the laboratory could be very material.

And let's talk about the cost savings, your last question..

And then honestly, that's a part of it. But at the end, the patient is also getting the insight that they need. I mean if it was someone in my family, I would want them to be seen by someone using the DDM platform based on the learnings that clinicians can have relative to the prevalence of that variant and some of the other conclusions comparative to the community, right? And even in a multimodal sense where you may see associations, right, relative to drugs than on a pure biomarker basis aren't correlative. But if you take 9 different data points from different modalities and together you find a correlation, you might actually much better solve for why we have responders and nonresponders for certain therapies.

And so it's a sell from you, I know it's a B2B, but like it really requires the physician because the physician right now in sequencing is still the biggest hurdle in the education standpoint. So how much of it is all right, we sold to the hospital. Now the hospital has to educate the physician, like you need to be using your patients here or -- for your patients, they're getting better care. Like how much -- like talk about that push?

Yes. This is probably one of the more challenging aspects, right? Because you would think like all these folks in the hospital talk to each other, they don't. Like the pathologist doesn't speak to the oncologist often and let alone their radiologists or others, right? And so this is one of the things. So certain customers of ours, one recent one solved it by mandating it. And they just said, look, we've made this decision organizationally. It came from a very high individual. We think this is better for us and our patients. Everyone needs to stop sending out and they need to now send the volume to our internal laboratory. Then we have other places where it's really tough for them, right? They're still sending a few thousand samples out a year. And their lab is running one of our solutions and the clinicians prefer the send out because I think it's lower risk or it's easier or there's like lab coverage or there's going to be better reimbursement, right? And so I would say, again, there's still a lot of education to happen. As we move into our CarePath module, and maybe you want to talk about that, as we start to touch the oncologist, I think that will become a bit easier.

Yes, sure. So CarePath is our multimodal analytics module that we launched late last year, and it's basically giving the clinician the ability to track patients longitudinally and do things like compare them to other patients, visualize multimodal data and then eventually predict outcomes and compare them to others in order to choose the best type of treatment.

Awesome. That's all the time we have.

Thank you, Luke.