Seer, Inc. (SEER) Earnings Call Transcript & Summary

Good afternoon, everyone. Welcome to the 2022 Bank of America Healthcare Conference, live from beautiful Las Vegas. Nice to see everybody here. I'm Derik De Bruin, the Senior Life Sciences and Diagnostic Tools Analyst for Bank of America. And now for something completely different, let's talk proteomics, here we are today with Seer. And with us from Seer is Chief Executive Officer and Founder, Omid Farokhzad, and CFO, David Horn. Gentlemen, thank you for being here. Thank you for coming in to be with us today.

So start off with the question and we were just chatting about this. You guys were the first proteomics company to go public, and then there was this wave of other companies that went public. And I think one of the questions I get from investors is, how do you sort of differentiate all these companies? So can you sort of put Seer in the competitive landscape and sort of would like to put some dots on the map and tell us sort of like how everybody fits together in the picture?

Yes. So first of all, Derik, thank you for inviting us. We really appreciate the opportunity. So I put the proteomic companies in 3 buckets. But before I do that, let me tell you what Seer is. Seer is a company focused on delivering unbiased or untargeted proteomic information at scale and speed that was previously not possible to do on by studies at any scale. The reason that's important is, that unlike genome, where we have 20,000 genes and all of our cells more or less have the same genes, and in the proteomic space, those 20,000 genes code for million or several million different protein variants. And these variants of proteins are important in health and disease. And we literally know the tip of the iceberg about this universe of protein variants, and to study them, you have to look at it in an unbiased, untargeted way. So that's where Seer comes in. So I look at the proteomic space, and I put them in 3 buckets. Bucket one is, folks that say we're going to look at the proteome in a targeted way, and the one that was -- the earliest one to be public would be, for example, Quanterix. Others are like SomaLogic and Olink, and they have different approaches to do that, but for the most part, they're all based on a ligand, could be any ligand, antibody, aptamer -- by the way, there's no shortage of ligand, affibodies, nanobodies, aptamers, I can go on. And that ligand interrogates a given protein that it was designed to bind to. Keep in mind that an average human protein is about 470 amino acid long. An average binding side of a ligand is about 5 to 8 amino-acid long. So when a ligand binds to a protein, it only sees that part that it binds to, but if the variant of the protein makes changes, anywhere else who doesn't see it. So that's a targeted approach, and so a targeted approach theoretically will max out at the 20,000 proteins, 20,000 genes, 20,000 proteins, you could theoretically end up with a 20,000 plex panel. And then there's another bucket of companies that say, we value unbiased proteomics. We think the problem is, the detector is the reason it didn't scale. And so we're going to -- and by the way, the unbiased detector today is the mass spec, and they say, we're going to replace the mass spec with a new detector, and then that companies -- or that bucket of companies includes Nautilus, Quantum-Si, on the public side; Erisyon, Encodia on the private side, and they basically got different platforms. They can look at the complex in the proteome. And then Seer is the one company that does unbiased proteomic and there's exactly one company that sits in that bucket. So we're not a detector company. In fact, if any of these detector companies eventually gets a product in the market, that product has commercial traction, then the Seer technologies, Seer product, which is the proteomics product, [indiscernible] that product, that would be a detector. We are detector agnostic. Our detector of choice today is a mass spec. There's 50,000 of them installed, growing at about 8% a year. About 1/3 of them do proteomic work. And about 1/3 of those 1/3 does deep unbiased proteomics, but the thing is they were doing it in a way that it was not scalable. So before Seer, the largest unbiased proteomic for example, was -- in plasma were studies that were in the tens of samples. In fact, the largest published study was 48 samples. But Seer now enabled studies to be done at the scale of thousands. In fact, we have customers that have already done thousands, and I expect that to actually scale to 10,000 and frankly, beyond that. So -- and we -- and I don't consider by the way, targeted approaches to compete with an untargeted approach, because they answer fundamentally different sort of questions than what an unbiased approach does. We believe just like genomics, an unbiased access to genomic content really opened up entire new end markets, liquid biopsy, NIPT, et cetera, we think unbiased access to proteomic content is going to expand existing end markets, open entirely new end markets, and that's how we see the difference, Derik.

Great. Thanks for the intro, that sets the stage. You just -- you've recently begun broad release of the Proteograph. I guess what's been your feedback from your early access customers and what are some of the applications you're using? What are some of the areas of improvement? Just what's the initial feedback on your initial product launches?

Feedback has been really great across the board from the instrument going to the customer, kind of hitting their labs to install time to validation, training. You basically go from an instrument, arriving to an instrument being validated for the customer in about a month's time. Training is relatively quick. They typically start up with a small study and then they scale to a larger study, batch-to-batch variability, play to play variability, those numbers have all been great. Fairly across these samples, those numbers have been extremely well. The software suite is easy to go from a biological content to biological insight. So there, overall, really great feedback, but we're early in the game, but it's been very positive.

So what's your -- what does the sales cycle look like? And like how much -- I mean, as you pointed out, there's only a fraction of labs that are doing unbiased proteomics. I saw some, [ it's mostly dose ] to the customers doing it, and sort of like what's the sales cycle and sort of how has that been impacted by COVID? I mean these are things you need to go in and people need to go in and feel the instrument, touch it, go in, and obviously, that's been difficult with the situation?

Yes. Look, I mean when you put an entirely new transformative product in the hands of the customer, the likes of which they have never had before. In other words, unbiased proteomic at this scale was never possible before. Literally, the largest published study was -- in plasma was 48 samples. We just had Oregon Health complete their 1,000 sample study. PrognomIQ is doing 1,000 plus, and we just signed up an aging one for 1,500. So when you put a totally transformative and something that people are not familiar with, neither the customer, and neither the scientist, you would be skeptical. You need to kind of understand it. So typically if I look at the early days there last year, and we just started a broad release, just 4 months ago, average customer who would want to do it, is well a concept study, this proof of principle studies, that are typically the tens of samples. By the way, ironically, our POP study in terms of samples, happen to be the size of the largest proteomic studies that were done in the past. As to a typical POP study happens very quickly, then they kind of analyze data, they kind of figure it out and then a customer will take that in. But my expectation is that, as data begins to emerge, and I don't mean our own data, like what we publish or what we present, I mean, customer data, what they publish and what they present, that over time, the need for POP studies will decrease. But in the early days, you need it, because scientists inherently are skeptical, and they should be, to make sure that the product does what it's supposed to do. I mean, we make big claims that you get thousands of proteins, the instrument runs 16 samples in 7 hours. I mean, people get to that depth of protein coverage, those studies would typically take weeks to months to do. So these POP studies are super helpful, Derik, in kind of getting customers over the fence. And then -- but from a CapEx perspective, the instrument is a couple of hundred thousand, most labs are easy to kind of grab, consumable, again, kind of price between an exome and a genome, if you would, in terms of pricing. So the labs are used to spending that kind of money. But the sales cycle, Derik, is long, because it's a new product, in the customer sense.

So I guess -- I mean, do you have any initial visibility on what pull-through would be -- what throughput, I mean, sort of like -- I mean, all these -- you have a box, there's reagents that go through it. Do you have any sort of sets on the metrics on, what it is going to be on a longer-term basis, or is it still too early?

Yes. I'll take that one, Derik, and let me reiterate our thanks for having us here. And so in terms of the pull through, again, it's pretty early, right, in terms of what we're seeing. Obviously, we've got a customer like PrognomIQ, on one hand that was -- the CEO is an ex-Seer founder, right? So no need for him in terms of convincing, right, and he's running thousands of samples. And so you can see that the pull-through there is pretty dramatic. I think the way to think about it though is, what's the typical size of study people are going to do. And what we've seen is that, typically, as Omid said, it takes about a month to install. And we expect it to take kind of 9 to 10 months after that to get to any kind of study of scale. And so they'll run the pilot study. So that will be tens of samples as well on their own to kind of, again, see what they're seeing, see what kind of studies they want to have. And then I think you're going to see studies -- you may see thousands, but you also may see a couple of hundred samples, right? So -- and I think that the hard part with these pull-through numbers is just their averages, right? So you're going to have people on the high end with a lot of samples and then people on the low end with only a few samples. So -- but we think typically, it'd probably be a couple of hundred studies -- a couple of hundred samples in a kind of a typical timeframe for that. But we're really too early right now to kind of being able to say we can notice any trend or any predictability, in terms of what the consumable pull-through is going to be.

And so the cost per -- the cost per study is what on average? And then obviously, I mean -- you want me to assume that that was -- that that cost would ultimately become -- move lower at some point?

Yes. Yes. So the cost, as Omid mentioned, the instrument sells a couple of hundred thousand dollars, which is typical of -- we base the platform on a fluid handling system from Hamilton, and so that's typically what they chose. We've custom designed it. We've got our own software interface on that to make it super easy to use and align with our panel and algorithm. And then the per sample cost is 2-tiered pricing. So for commercial, it's in the high hundreds of dollars per sample, so closer to a genome, right and then for academic, we -- it's probably mid hundreds of dollars per sample, that we're charging them. And again, the reason for the discount is, they are more typical -- you're going to publish, right? They're going to put their papers and presentations out there. Obviously, they have a more constricted funding environment as well. So just trying to give them the benefit of that and keep the discounts. Obviously, we will offer volume discounts to people who want to do sizable studies.

So speaking of -- I mean, obviously, you're new in -- you are new in -- it's a new product, people are getting it with it. So what does sort of like the publication cadence look like from your customers? And we have got ASMS coming up and some other sort of some other -- there are some other meetings sort of like later this year. It's like, how can we sort of think about the publication point? And I've never been able to figure out like, what's the tipping point when people say, we've -- aha, we have to have this product?

So Derik, so Seer published its first paper with the platform in Nature Communication. That was a paper published in 2020. Most recently, the company published another paper in PNAS, that was a 2022 publication. If I look back last year we -- and including the beginning of this year, a total of 40 presentations have been made in conferences. By the way, many of those were actually from our collaborators and partners and customers, and not just us. We're seeing the ratio appropriately so, as the customer base is slightly increasing than its early days, that presentations are more and more coming from the customers. So for example, at the upcoming ASMS, there'll be 14 presentations, 5 of which are customer base, 9 of which are Seer-based. And I think -- and the European Society of Human Genetics is coming up, we'll have presentations there, and customer present there as well. And so as the customer base increases, and the experience -- the platform increases, presentations and eventually publications will be largely from them and not from us. We will always present and always publish, but the large body will be customer-based and not based on us. Keep in mind, though, that the limited release, which finished last year, was largely based on the second half of 2021. And so an average customer kind of gets to a study of some scale, as David said, in about 9 months or so. So most of those limited release customers are going to begin to kind of have data now and then in the second half of 2022. And then the broad release customers that are now starting, they're going to generate data that's going to be the second half of 2022 and probably trickle into 2023. So as customer experience broadens, we're going to see more and more from presentation from these customers. And the thing is that at the end of the day, Derik, the scientific community needs these presentations, and the beauty of it is that science doesn't lie and science is awesome. And so the data that's being generated is extremely gratifying from industry leading scientists, really awesome data, depth of coverage. If I look at just the last AACR in April, PrognomIQ presented their study, 212 subjects, cancer study. Across the study, they identified 5,000 proteins and hundreds of glycoproteins. And these -- and so for any given protein, they would identify variants of the proteins as well. And so now you're beginning to look not just that, that protein is there, but like the variance of the proteins in health and disease, I mean, just that level of information, has not been possible before. And we saw that -- by the way, I expect that paper to be an incredible paper from the PrognomIQ guys. And then Oregon just finished their 1,000 sample study on prostate cancer. So I can't wait to see that data come out. So Derik, it's just going to take time for these studies to be done and they get presented.

And what about your partnerships with Thermo, Bruker, Danaher, some other mass spec vendors? How are those proceeding? And are they helping to co-promote the product at all?

Let David take that, and also maybe comment on the deal as part of it as well, actually.

Sure. So yes, so as Omid had mentioned, we are mass spec agnostic. In fact, we work with all mass specs. And so beginning of last year, as you mentioned, Derik, we did do partnerships with Thermo, Bruker and SCIEX. And really, those were -- in a sense, marketing agreements and that we would -- because what we wanted to do is, go to our customers and provide an end-to-end workflow, right? So we book in the mass spec. We have the upstream nanoparticle technology that helps with the sample prep to go into the mass spec and then we have the software suite on the back end that analyzes all the data that comes out of the mass spec. And so for us, being able to be agnostic and all the various mass specs have their pros and cons, depending on what kind of application you want to use it for. So we wanted to be agnostic to the customer, which we are. And the reason they wanted to do it with us, is because we are really enabling labs that in the past, would never even have thought about doing proteomics, to be able to do proteomics. And specifically, not only folks who've done some proteomic work that have maybe outsourced it, but folks on the genomic side, who have done these large-scale studies, but simply don't have the proteomics data that they want, to annotate that genomics data. And so for these mass spec vendors, they really wanted to access to that genomics market, and we are helping them do that. They are not promoting our product, but we are sharing workflows, we're sharing leads and things like that, but what we are doing is providing an introduction, providing some insight into the various options for our customers and then handing that off. And as Omid mentioned, we did announce a proteogenomics consortium in the beginning of the year with SCIEX and DLS. And these were -- again, DLS was a collaboration customer. They're one of the largest genomics service providers to big pharma. And along with SCIEX, who obviously, was one of our mass spec partners. And so what we announced there was, that there would be a proteogenomics consortium formed with the 3 of us to build the capacity for 100,000 proteomes per year to be able to be done. And of course, that's going to need to be scaled over the next few years. But it's been great working with them. Both parties are super excited about this. And really, the motivation is to just, again, impedance match, right? There have been over 1 million genomes sequenced, over 10 million exomes. But really just to give the capacity to a lot of these genomics folks, the big pharmas and others who have the genomic data, but they then want to get the proteomics data as well. And so who better to go into partnership with, than a very large genomic service provider, who wants to add proteins to their portfolio, and as well as SCIEX who, again, was very interested in accessing a market that they didn't have access to previously.

So one of the questions we always get when we're looking at these new technologies is, how big is the market? How big is the TAM? And that's always -- well, how big do you need it to be. Sort of the answer that we often give is like, well, do you -- how many genomes can dance on to head of a pin or how many proteins can dance on the head of a pin, right? It's those sorts of questions. How do you sort of size the market opportunity? Because I mean you can come up with some -- I mean, pick a number, any number really together.

Do you want to take that?

I'll take a start and then...

I'll add to that.

So look, Derik, it is large, right? And I think the way to think about it is, not only where it is today, but where it could be, right? I mean if you had asked, I'm sure, Jay Flatley this question in 2007, you might have gotten a similar answer, which is it's big, but ultimately, I think it's going to get really big, right, in terms of just the whole ecosystem that gets created around an enabling technology like this. So as Omid said, even just today, even if you just think small, which is there's 15,000 mass specs out there doing proteomics, you could set a Proteograph in probably every single one of those, right? And then the consumable stream that follows on. But again, I think that's a very limited way to think about it, in the sense that -- we are -- there's a whole market out there, i.e., the large genomic scale folks that we're going to be able to get access to this technology, which is not in that 15,000 number, right? It's just a whole new greenfield market. So again, I think it is a very large market, and it will enable a whole ecosystem, right? No one imagined NIPT or early cancer detection back in '07, right? It just wasn't really on people's radar screen. So we do think this kind of unbiased technology will spawn a whole new industries and a whole ecosystem around it. We're seeing a ton of interest in the technology from folks who are involved in neurodegenerative disease and aging, right? And just the role that proteins play in that, which is, again, an enormous market, you think about Alzheimer's and Parkinson's, and just aging in general, wellness. And so it's those kind of complex diseases that you haven't seen, a genomic signature be effective at all, really, quite frankly, that proteins will allow.

Maybe let me add a couple of points. One, just unequivocally, I think that proteins are the more informative analyte. So unequivocally, the proteomic market, the information that's going to come from it, in any way, research, clinical and not just human ag bio, microbiome, others, is going to be more informative, more actionable than genomic content is. And therefore, if you execute, that's a big IF, you're going after a larger TAM than the genomic TAM was. In fact, I think just about everything that genomic TAM -- proteomic will touch. There'll be things that proteomic will touch, that genomic was explored for, that did not work out, that's the first point. Second is, you mentioned, Jay Flatley, if I kind of go back in time, 15 years ago, and by the way, let's exclude the last 6 months of market pain, there was about maybe $20 billion of market cap in the whole genomic space in total. Fast forward 6 months ago, not today, that was about now $250 billion of market cap in the genomics. And the thing is that $250 billion, about $50 billion of it was the company's -- $50 billion, $60 billion of it, companies that generated the content in an unbiased way, because we didn't have content, right? Remember, we've now done 1 million genomes and 1 million exomes, that is content. So the companies that generated content, also cut into the applications. Those -- that was $50 billion, $60 billion. And then the other, let's call it, a couple of hundred billion, were application companies that came -- that existed only because those content existed. 6, 7 months ago, we had about $10 billion, $15 billion of market cap in the proteomic space. My prediction for you is, this turmoil aside, that 10, 15 years from now, that's going to be the same couple of hundred billion dollar market cap. And that a good chunk of that is going to go to companies that generated the content, probably a majority of it will go to all the applications that were enabled because of that content. So Derik, my prediction is that the proteomic TAM will actually be larger than the genomic TAM, time will tell.

So David, how do we think about -- if we get to like -- we get to like -- the angel has started singing. The -- if you start thinking about what the model looks like, when you're at scale and you're sort of running, so let's take it that you are $1 billion in sales, right? Nice round number.

Nice number. I like it.

Right. Say you're $1 billion in sales or somewhere at that point, it's like what's the model look like, right? What's the margin? What's the right OpEx structure?

Yes. Look, it's an extraordinarily profitable model, Derik. We have -- in the long term. We have very high gross margins on the kits, nanoparticles and reagents that go along with it. This is very high-margin consumables, and then an instrument that is not high margin today, but we feel that there are opportunities to work on that. And so we've said publicly that the gross margins long term, we've been in the 70% to 75% area, and that operating margins in the neighborhood of 35%-ish. So what you would expect to see from other very high -- very highly profitable life science tools.

Any questions from the audience? As we're sort of winding down on time here, my standard ending question is what do you think is underappreciated about Seer?

Underappreciated? I would say probably what is most misunderstood, that's the most important point, is this whole concept of plex. For example, I have 1,500 plex or 3,000 plex or 6,000 plex, in terms of a targeted approach. There is a very different way of thinking of it in Seer. As when they say, Omid, your product is 5 nanoparticle. Is that 5 plex? And we say, well, that's a totally different way that you should be thinking of it. The reason is that, an average gene creates about 100-plus protein variants. Different body compartments express different genes and different protein variants. If I look at across our studies, we actually get to -- and I think we mentioned this at the J.P. Morgan conference, a 12,000 protein variant, -- proteins -- actually protein groups, not protein variants, protein groups, across the different tissue types. And so when the numbers we give is in the plasma, for example, because that's where our focus is. But slice it in others. So you can't think of it, in terms of plex. The other important part of it is this. Today, we have 1 billion genetic variants that have been identified. The very first person we sequenced, did not have 1 billion different genetic variants. As you do larger and larger studies, you create -- we generate more and more of this variance. So with SEER study, as you begin to do studies of scale, we will actually identify using the Seer platform, tens of thousands, then hundreds of thousands, then hopefully 1 million-plus variants of proteins. So in that context, just thinking of it in terms of plex is wrong, and I think that is probably the one thing, Derik, that I want to get across very clearly.

And you're right on schedule. Thank you gentleman. Thank you for being here. Thank audience for listening. Appreciate it. Thank you very much.

Thank you so much Derik. Really appreciate it.