Hey, everyone. Good morning. I'm Tejas Savant, and I cover the Life Science Tools and Diagnostics sector here at Morgan Stanley. Before we begin, important disclosures, please see the Morgan Stanley Research Disclosure website at morganstanley.com/researchdisclosures. And if you have any questions, do reach out to your sales rep. It's my pleasure this morning to host Seer, and on behalf of the company, we have Omid Farokhzad, CEO, and David Horn, CFO. Thanks, gents, for spending the next 30 minutes with me. Omid, I want to start, you know, at a high level, right? Despite some of the near-term headwinds which, you know, we'll get to later in the chat, I want to talk to you about sort of what's gone right over the last 12 months for Seer.
At an industry level, you know, the excitement around multi-omics in general, and proteomics specifically is palpable. What are you most proud of in terms of what you've accomplished since IPO?
Okay. Thanks, Tejas. Number one, we built a great team, and, and I'm super proud of the group of people, like-minded, that came together to execute. Number two, we aim to put a product together that could give us access to the proteome in an untargeted, unbiased, or hypothesis-free way, in ways that would open up a new gateway to the proteome. It's important for context that before that instrument was put in the hands of the customer, if you try to do proteomics in an unbiased way, deeply, meaning you, you begin to see low concentration proteins in the biological samples, that you are limited. You have to make... You have to compromise between how deep you want it to go because the workflows were tough. And then what type of studies you could do, the studies were all underpowered.
Biomarker discovery was a challenge. For context, the deepest plasma proteomic study published before our instrument went out the door was a study with 48 samples, the biggest. The deepest study, meaning the number of proteins that they could identify, was 4,500 proteins that were studied out of the Broad. Okay, so then we shipped the first instrument. By the way, as of the end of last quarter, an average customer has had our instrument in their hand just about one year, 3.9 quarters, to be exact. So most customers have had, on the average, about a year of the instrument use. So what are we seeing? Number one, we've now had multiple customers complete studies of 1,000+. One customer, PrognomiQ, starting studies of 15,000. Depth of protein coverage in plasma has now reached 8,000.
At the ASMS Conference in June, when a federal official launched their new mass spec, the Orbitrap Astral, they actually exemplified the performance of that instrument with the Proteograph. And they showed in their own data that they get to 6,000 protein with 45 minutes per sample mass spec time. I mean, these are just levels of performance that were fundamentally not possible, highly disruptive in what's possible. So I'm very proud of the way the instrument is performing in the hands of customer, how easy it is to use, and the data that's being generated. And, you know, but we're early in this game and the path from where we are to really putting proteome in an unbiased way to have the customer, the way access is available in genomics, that path isn't going to be a straight shot.
But we're plowing through it, and I'm very optimistic about our opportunities ahead.
Got it. So on that last point, Omid, on the last earnings call, you talked about, you know, sort of the three factors that represent, you know, the key barriers to adoption for the Proteograph. Maybe just starting with access to mass spec, right? That's probably, you know, the most important one. You've identified that as one of the key, you know, barriers. Can you remind us what the install base for mass spec worldwide looks like, and what fraction of this install base are using mass spec for biological use cases?
Yeah. In install base, I don't have the updated information as of now, but, but the number, a couple of years ago were 50,000, growing at an 8% CAGR, 50,000 globally, growing at an 8% CAGR, and about 1/3 of them are used for proteomic applications. Within that segment that are used for proteomic applications, about a third of that are kind of involved in the. If you would try to see it, the tinkering that goes on to kind of try to get workflows that goes deep into the proteome, for various different applications, whether it's plasma or tissue or cells, et cetera. That's the mass spec target audience.
Mm-hmm.
Then, if I look at what does that mean for us, and what have we learned in this process? We had always expected that our customer base and our term would be a combination of genomic and proteomic. In fact, early on, right after the IPO, if you remember, we did partnership with Thermo Fisher, with Bruker, with SCIEX, because our value proposition to, to, to those providers was that we expect to put the Proteograph in the hands of genomic customers that are not mass spec users. And in fact, them accessing our instrument will open up entire market for you, that you don't even tap into today, in addition to the traditional proteomic market that you're in. So we had identified those two customer types from the very, very get-go. But what did we learn as we went to the market?
Here's what we learned. Number one, the proteomic customer. This is a mass spec user, that does proteomic work because they could never scale.
Mm-hmm.
As I mentioned, the first, the biggest study was 48 samples, the deepest was 4,500 proteins. And by the way, in that study of 4,500 proteins, that was a 16-sample study. Because they could never scale, they don't have freezers full of samples. Also, their business model, in large part, is that they would invest a lot of money on CapEx, and then they didn't really have a consumable business, so they weren't used to this OpEx continuous paying for samples. And the way they worked was they had FTE costs and those people's kind of did fractionation, depletion, these things for time. But that's the way those labs operated, without samples, really.
So now here we come. And by the way, the uniqueness of each of those labs and the publications that would come from it was the very unique nature of their workflow, and they all differed from each other. So and so would be this method of depletion and that method of fractionation, and somebody else does it differently, and they would go deeper and deeper or, you know, or do it in a different way. Now, here comes here with an instrument that takes the black art out, standardizes the process, automates it, and gets to depth, throughput, and precision and reproducibility that is totally disruptive.
Mm-hmm.
And takes away all of those uniqueness of these different workflows and just fundamentally standardizes it. Okay. Now, but those folks don't really have samples, because now you're letting a mass spec user that the biggest study would do would be in the tens of samples. So okay, now you're ready to go with thousands of samples. They just don't have freezers full of samples. Now, that same lab that did the deepest paper, they just presented data at the ASMS. Their deepest paper that was published in 2017 was 16 samples, 4,500 proteins. Now, they did a 300-sample study going to a depth of 6,000 proteins. For them, going from 16 to 300 is a big step up, but we're actually interested in customers that can do thousands or tens of thousands of samples.
We wanna see papers get published equivalent to the UKB Nature paper, where 450,000 exomes were sequenced.
Mm.
That 450,000 deep unbiased proteomics got done. With the latest product launch, the Proteograph XT, you really do now are able to impedance match the throughput by which we access proteome and genome with CapEx that is modest, that most organizations that are scaled in that way can actually tolerate and do. Okay, so then what about the genomic customer? The genomic customer actually, doesn't really care about what the protocol is, whether it's a mass spec or not. Mass spec is complicated for them. They're users of biological data. They just want the data.
Mm.
They don't want the complexity of, you know, I have this limit, the amount of column chromatography time or this. They just want what is the biological insight that comes. Given those two different customer segments and what we learned in the market, I think commercially, we've now adjusted. I'm very proud of the team in the way they saw the opportunity and the shortcomings of the way we were viewing it, but also the differences in their viewpoints with their constraints and addressing them. The first, we were always adamant about wanting to have a distributed commercial model where we would place instruments and consumables, and we wanted the instruments everywhere. We did not want to be in the service business. We still don't, by the way, Tejas.
Mm.
That said, in order to catalyze the community, we need peer-reviewed publications. So you gotta fill the top of the funnel with as much data as possible, so on the bottom, papers come out.
Mm.
Because that is so important. In order to fill the top of the funnel with data, 'cause the data that's coming out, that I have visibility to, and as a customer, it's just fantastic. But to fill the top of the funnel more, we started the STAC, which is a Seer Technology Access Center. The STAC essentially does two things: One, it gives end-to-end solution to a genomic customer that typically does work on a service basis. And they send us a sample, we send them data back. Simple as that. But also, for the same genomic customer, we provide what is called an XTM service. Our latest product release was Proteograph XT.
Mm-hmm.
The XTM is that we sell a kit to a customer, and that kit, the price point of that kit is designed that the customer can have a Proteograph, they can run their own Proteograph, but as part of that price, they send the peptides back to Seer, and we run the mass spec for them.
Mm-hmm.
Again, this is all to lower the barrier for the genomic person to get hooked on the data.
Right.
Because once you understand the value proposition of the content that comes from the proteome in an unbiased way, I think necessity is the mother of all invention. Once the genomic customer knows what that means for them, they will figure out a way to bring a mass spec in, or the mass spec guys will figure out a way to provide that for the genomics at the right level, and we wanna facilitate that. For the mass spec guy that doesn't really have a lot of the samples, and we want them to engage. By the way, that applies to both customer types. We also introduced the SIP, and I will have David comment on it.
Mm-hmm.
And the SIP, essentially the strategic placement, of the instrument, where we place the instrument, but it's tied to consumable purchase upfront. We screen folks that are able to run samples. The motivation is that you gotta publish your studies. And then those customers are screened in order to be able to buy the instrument later. But it's really, again, to lower the barrier. All of that said, Tejas, every piece of evidence, by that I mean scientific data, that I see coming from any customer that I use the Proteograph, totally makes me feel great about what is possible once access to unbiased proteomics becomes readily prevalent.
My prediction for you, and I made the same prediction a couple of months ago, is that within 18 months, so by that I mean, by the way, a couple of months has gone by since the last time I made a prediction, so I'll stick to it. Within 15 months, by the end of 2024, I predict the first population-scale, deep, unbiased proteomic study will either be done or at least well underway.
Got it.
I think the content value of that will be unprecedented.
Got it. I mean, we love forward-looking statements, I mean, so.
You can keep my feet to the fire there.
I guess the question is, how confident are you that it'll be on the Proteograph? And, you know, as a follow-up to that, I mean, you had, you know, user presentations here, actually, at ASMS a little while ago. And we got a lot of follow-up out of that around, like, you know, could someone like a Steve Carr perhaps, like, run a bunch of, you know, UK Biobank scale, you know, samples or perhaps samples from the UK Biobank on the Proteograph?
Yeah. We're seeing an enormous amount of innovation on the mass spec side. If you look at the Orbitrap Astral or even the Bruker HT, I mean, these instruments are doing so much more than they could before. But if you just look at the Thermo Fisher data that was presented at the ASMS, the Astral, when you do a direct injection of something like plasma, so what would, for example, come from a UKB sample, they get to a depth of 675 proteins in their own study. Now, 6,000 proteins in plasma on a single injection of mass spec is phenomenal. It's just so different than three, four years ago if you would look at unbiased proteomics.
But then the same instrument, you put a Proteograph XT in front of it, and suddenly the plasma produces 6,000 proteins.
Mm-hmm.
We now have customers that have used a combination of those two and are getting to 8,000 proteins because as the sample size increases, content also increases. So the thing is, I don't see a path today to a commercially viable, scalable solution, none that I have any visibility to today, nor if I look over my shoulder and I look backward, that can do a study like deep unbiased proteomics, that does not involve the Proteograph.
Got it. That's great. So, I mean, I want to spend a little bit of time on XT. I mean, obviously, you've mentioned it a couple of times already. Can you just give us a, an overview of, you know, how the assay improves upon the performance of your original Proteograph suite, and just feedback you've heard from early adopters to date?
Yeah. When we launched the first product, the Proteograph Assay, it involved five injections per samples. It's because every sample was split into five wells. Those wells had our proprietary engineered nanoparticles in them, and then those five wells each went into a mass spec and a separate injection at 30 minutes, each total time was 2.5 hours of injection time. That was a very significant step up to what was possible before in terms of depth and throughput. What the XT does is it further increases the throughput by more than 2x, 2.5x, to be exact. You go from five wells per injection down to two, and so it's two injections per sample type. Now, the...
If you look at the Orbitrap Astral data that I shared with you, the 6,000 protein, that is actually done on 22 minutes per injection, so the total sample injection time is 45 minutes.
Mm-hmm.
Right? So one Astral and one Proteograph together now can generate 10,000 samples, deep unbiased proteomics per year. That is not a big commitment in terms of infrastructure to the large scale. What that means is, if you now want to run 50,000 samples, you know, that's five times of that.
Mm-hmm.
We're definitely getting to a point where you should be able to do large scale, population scale studies deep with that number of proteins, the kinds of procedures that just are phenomenal.
Mm-hmm.
It gives you so much power in terms of discovery for biomarker changes in proteomic signatures. And so that is how the XT operated. It reduced the throughput, preserving the content and preserving the precision of the assay.
Got it. Makes sense. Do you think the price tag of the Astral is a bit of a gating factor to seeing that elasticity of demand play out on the consumable side for you guys?
I think the Astral is a very expensive instrument. But I think the performance of the Astral is also highly differentiated. Obviously, we have Astrals in our lab. We have three of them in our lab. And you know, my expectation is that the value of the content will drive demand. I mean, I know genomic prices are down, and we've now sequenced 1 million genomes, 10 million exomes, but you guys have all been around the block for a long time. Just look back a few years. It cost 10 x the amount to do a genome sequencing, an exome sequencing, and people yet paid because the value of that content at the time was appropriate. I think that we don't need to chase the price down.
We need to demonstrate the value of the content. I think the value of the content will speak for itself. Now, that said, until the value becomes clear, it is a bit of a price war. But I think once you're differentiated, then I think it stands on its own two feet. Now, for a customer that doesn't want to spend the money on an Astral or Orbitrap Astral, the Bruker HT, that's also an exceptional performance at a much lower price point.
Mm-hmm. Got it. I want to talk about, you know, the strategy for getting more publications out there, 'cause that's clearly, you know, one of the gating factors to broader adoption. You know, collaborations are something that a lot of companies use to spark these publications from key opinion leaders. Can you describe how you utilize that strategy? And what is the cadence of publications that you're expecting between now and, perhaps the next, six to nine months or so?
Yeah. Tejas, it's always, I mean, having spent most of my life in the academic world, where your life is writing grants and writing papers, it's always difficult to go through a peer review process from a publication perspective. I mean, I'm not doing it now, because I'm now CEO of a public company, but our customers are doing it, that are trying to publish this. There's an inverse correlation, usually, other than the COVID time, with regards to, speed to publish and the quality of journals. If you want to go to the top journals, the peer review process takes longer, review can be multiple cycles. If you go to a lower quality journal, you can get things published very quickly.
Some of our customers that are now in the manuscript submission process or review process, some of whom, Tejas, you know, have targeted very, very high-level journals, and by the way, appropriately so, because the types of content that they're trying to publish was never possible, and so they're aiming high. You can imagine, you know, those top one, two journals in the scientific community. We have a customer that's now in the revision phase there. We have another customer on another one of the sister journal of the same journal that just got reviews back from a reviewer that is very positive, and so I expect that paper to be published.
But the cycle is long, and from my perspective, I sit here pulling my hair, waiting for these things to come, because every one of them that I'm aware of the data is just fantastic content.
Mm-hmm.
It's just the process is long. Now, keep in mind, though, as I said, that the average customer has had a Proteograph products within their hand in exactly 3.9 quarters, one year.
Mm-hmm.
It takes time to do a study, to then go through a peer review process and publish it. We're early.
Mm-hmm.
We have not been around as an organization commercially since 2000 or 2004. We shipped our first instrument, you know, two years and eight months ago, and an average instrument has our customer a year. But what's coming, and the visibility I have to the content is great. My expectation is that we'll begin to see at least two or three publications from customers in peer review publications this year. And then the visibility I have to the pipeline is about 11 or 12 papers that are in various stages.
Mm-hmm.
So I think those will then trickle into 2024 in terms of when they'll come out.
Got it. Is there a possibility that we might see some of those in preprint form before they make it out?
Yeah. There's actually, actually, Tejas, there's about two of them in bioRxiv. One looked at, for example, other than biofluids or liquid samples, looked at skeletal muscle.
Mm-hmm.
Tissue and looked at tissue with a Proteograph. That's interesting because remember, what the Proteograph does is it compresses the dynamic range, but it's quantitative. In tissues like muscle, where most of your proteins are one protein or two proteins, which is like, you know, like myosin. But then and that to be able to decrease the compression of the dynamic range is the lower bound of protein, Proteograph adds a lot of value for that. So that paper is looking at amplification of tissue.
Mm-hmm.
S keletal muscle. That's a customer paper from Auburn University that's now on bioRxiv, and that paper is under review. And then, Karsten Suhre' s paper, which is looking at pQTL, is on bioRxiv. That paper, by the way, their review came back. That was the one that I alluded to. That looks really good, so I expect that paper to come out in a peer-reviewed form, hopefully by the end of the year.
Got it. Got it. Switching to macro. I mean, obviously, we talked about sort of the cautious, you know, spending from customers, and you mentioned SIP as one of the initiatives that you've launched in response to that, to avoid, you know, the upfront CapEx involved. Tell us more about, you know, how you go about sort of selecting appropriate sites to be, you know, part of SIP. And, you know, how do you see the mix between SIP placements and straight-up placements evolving over the course of the next 12 months?
Yeah. Tejas, thanks for that. In terms of the SIP, you're exactly right. What we, what we try to do is look at kind of these key opinion leaders and folks that we think have projects in near term, because, again, getting back to the... We're trying to put as much data in the top of the funnel for folks. And just given that some of the, both the academic folks and increasingly, you know, some of the commercial folks, have constraints on their capital budgets, at least, you know, this year, as they think about budgeting for next year, or they're in the grant cycle, still in the- in that process. What we've said is, you know, purchase consumables upfront.
So if you purchase a certain amount of consumables upfront, we will then loan the instrument for a period of time, you know, up to 12 months. And then, you know, they have projects that we expect the consumable pull-through to work through with that instrument. And then, you know, certainly, as Omid said, we expect they have the intent to purchase that instrument at the end.
Mm-hmm.
So really, it's a budget cycle-type loan program where you're able to place the instrument. Therefore, they can get started on those projects.
Right. Mm-hmm.
And start to produce that data, and then, you know, work through the budget process on their end, to ultimately purchase the instrument.
Got it. And are you starting to get sort of customer inbounds interested in looking at SIP and accessing the platform that way?
Yeah. We've been pretty successful with that in terms of there's been good uptake .
Mm-hmm.
Around that. I would say certainly we've seen some traction in that. You know, we really kind of launched it in the first quarter, so we've seen you know, quite a number of placements under the SIP program.
Got it. Do you... Is the consumable package that, you know, customers have to commit to before they participate, and is that standard, or does it sort of depend by customer, their circumstances, and how large of a commitment they were, they are willing to make?
Yeah. That generally we kind of work with them to determine, you know, what size of project they're looking at-
Mm-hmm.
What types of commitment. There is a threshold under which we won't go below.
Got it.
Right? So there's a minimum that we would consider it. And so, certainly that's considerations as we look as we evaluate the customer.
Got it. One of the questions we've gotten, David, you know, following the launch of SIP is just, passing, you know, the SIP placements out from straight-up placements, right?
Mm-hmm.
And to your point, I mean, over the long term, it doesn't matter because it's the install base, and that drives consumable growth. But at least for, you know, the next year or two, would you consider sort of separating those out when you update the install base at the end of the year?
Yeah. I think we'll certainly give you some qualitative feedback.
Mm-hmm.
Around what that is. And I can tell you now, Tejas, it's around, it's been around half.
Got it.
In terms of, this year, in terms of what we're looking at.
Got it.
I can...
Got it. Switching to China, I mean, obviously, you know, a topic of great interest at this conference, and varying levels of bearishness around it, you know, around some people calling it sort of a structural reset, others saying it's more of a temporary thing. Talk to us about, you know, what you're seeing, you know, in China in July and August, Omid.
It's been soft, Tejas. It's been soft, little activity, certainly a lot more than we had expected.
Mm-hmm.
I don't have visibility to that actually changing the balance of the year for us either.
Got it.
David, you want to add to that?
No, again, I think we certainly expected more, and again, we just haven't seen the demand come through China.
Well, just in terms of the conversations you've had there, like, is the sense that this reset is here to stay, and then 2024 might see a continuation of the trend, or is it more a situation of, you know, you don't know where the bottom will be over the next, you know, two quarters or it'll be three quarters?
Yeah. Look, I don't think we... That's the part of the frustrating part, though.
Mm.
You know, with our partner, Enlight over there, we do have discussions with them, but, you know, they don't have a ton of visibility. They, they're optimistic.
Mm-hmm.
But, so that, you know, hopefully next year is more encouraging. But that's part of the frustration on our part, too, is just a lack of visibility.
Got it.
That when does it turn? And again, we're small, right?
Right.
Um, but-
Right.
B ut just from what we've seen, again, a lot of, you know, conversations going on, but unclear as to when the funds will free up for that.
And just, you know, more broadly at a global level, Omid, you know, biopharma versus academic, how are the trends looking there? I mean, the one hand, you have, you know, the, the funding headwinds, and the other hand, you know, this year seems to be okay, but there are some concerns around, you know, potentially a little bit single-digit decline in academic budgets next year. So just walk us through, you know, what you, what you see in both of those end markets.
Yeah. The pipeline in terms of demand is almost split in the middle.
Mm.
If I look at sales, it's probably 2/3. 1/3 commercial, academic.
Mm-hmm.
Then if I look at pull-through, very strongly toward commercial versus academic. Keep one thing in mind, which is academic entities usually run on grants.
Mm-hmm.
When you submit grants, usually you cite references and papers in to support your methodology and approaches. There's just a paucity of data-
Mm,
Today to really support even an academic grant process. But as David pointed, we have placed instruments in several high-level academic labs. One of the strategic changes we made is that there are academic labs that are also largely supported by philanthropic approach, you know, sources.
Mm-hmm.
We're targeting some of those. By the way, they just happen to also be kind of like genomically oriented academic labs. And so we are now beginning to place, and there is one very significant one that is happening in this quarter placement into labs that are genomically focused academic labs that are heavily stacked with samples, very used to running very large-scale studies, and their funding isn't your traditional, you know, kind of government grant funding.
Right. Right. Got it. You know, to finish on an optimistic note, right? So you've talked in the past, Omid, about increasing the content on the platform, you know, reducing sample input volume and throughput. So when can we expect to hear, you know, improvements on the other two dimensions?
Yeah. Tejas, the sample volume part of it is actually something that we're working on. In fact, it, you know, I was in the therapeutic space for most of my life, and in the therapeutic space, it's almost like a label extension, if you would, in terms of once you get a drug approved, you then get it for other indications. Even for XT, our apps lab has exemplified the use of that with lower volume. So we're gonna do lower volume, not as a new product per se. I think it's too big of a... For me, a product needs to be really radically different. So sample volume to me is an exemplification that we're doing on apps lab. You should expect that in the XT. It's already happened.
We'll write white paper on it, and we'll support customers to lower volume on the XT. I think you can also expect from us that we'll do about a cadence of a product a year.
Mm-hmm.
For me, a product is, could be, by the way, an instrument, could be a new kit, could be software. But if it's software, it needs to be radically different.
Mm-hmm.
Like, for example, on the 18th of this month, we'll do a software update. It's a pretty important update, but I don't consider it a new product.
Mm-hmm.
So if it's a software, it'll be dramatic. If it's a hardware, it'll be something very different. Just to give you context, we're now doing hardware upgrades for the XT, but I don't consider that a new product. So for me to call something a new product, the magnitude of the impact it needs to have, has to be as different, for example, as the XT was.
Mm-hmm.
You can expect about one of those per year. I think it's also fair to say that the next product will be kind of a content-focused product.
Got it.
Yeah.
Fair enough. We're out of time, so that's a great place to leave it at. Thank you so much.
Thank you so much, Tejas. Really appreciate it.