In this episode of Founded and Funded, Madrona Partner, Chris Picardo, sits down with Dr. Ali Ansary, founder & CEO of Ozette. Chris and Ali talk about Ozette’s mission to uncover hidden information about the immune system to help with the efficacy of cancer treatments. Ali also speaks about the ability of biotech companies to attract traditional tech talent.
Transcript
In this week’s founded and funded Madrona partner. Chris Picardo speaks with founder, Ali Ansary about Ozette, Ozette is what we call an intersection of innovation company. Blending innovation in life sciences with machine learning and AI to uncover data not available through conventional means.
Ozette is focusing on the immune system and specifically how various cancer treatments interact with the immune system. As therapies for treatment, get more complex and doctors are evaluating how to combine various treatments, especially in the case of cancer treatments. Data like this. Derive from Ozette can have a direct bearing on a patient’s experience.
Ozette spun out of work and an open source project. Developed at the fred Hutchinson cancer research center. [00:01:00] Combined with incubation at the Allen Institute for artificial intelligence. Chris and Ali talk about building a multidisciplinary team. And the unique. Attraction that a life sciences based.
Company has for traditional technical talent.
Is a great conversation. Listen on.
[00:01:23] Chris Picardo: Ali super excited to have you here on the Madrona podcast. And it’s always fun to have these conversations, especially with entrepreneurs who are early in the journey and learning new things every day. So excited to have you join us. Ali is the CEO of Ozette which is a recent investment that Madrona made and was spun out of both the Fred Hutch and the Allen Institute incubator here in Seattle and we’re incredibly excited to partner on the journey. One thing that’s so interesting about Ozette is that this is another investment that Madrona has made and what we’re calling the intersections of [00:02:00] innovation, which is companies that are operating at sort of this intersection point between cutting a wet lab, life science techniques and machine learning and software. We are super optimistic about the future of companies who are working there and Ali will certainly share more of the reasons why we’re so optimistic. And, you know, we’re excited to have this discussion.
So without further ado, I’ll introduce Ali he’s the CEO of Ozette. Ozette is a immune intelligence company that combines the best of single cell analytics with cutting edge machine learning. When we say single cell analysis, we’re talking about the analysis of individual cells in a patient’s sample. It’s often used to diagnose conditions like leukemia, and there are many other emerging use cases and Ali will certainly talk about some of those.
Ali, that’s a lot of me talking. I think it would be great for you to spend a little bit of time on the background of Ozette and how it was founded, how you originally got [00:03:00] involved and where you guys are now.
[00:03:02] Ali Ansary: Well, I appreciate, Chris, you’re having me. I’ve been looking forward to spending a little time, I know you and I speak on a weekly basis, so I think it was a natural time for us to put this down into recording. I appreciate it the Madrona team has been absolutely supportive and like you had mentioned, you know, we we’ve been fortunate that, Ozette’s being built in a, what I would probably say the capital of humanology. A lot of it actually predates, Ozette and the work that had been pioneered at the Fred Hutch, with E. Donnall Thomas who had received his Nobel prize in pioneering work in bone marrow transplantation in the early nineties.
That basically put a springboard for the Hutch becoming a pioneer in cancer research, infectious disease. You had Immunex who was known for their therapy for Enbrel which Amgen had acquired. So, naturally what has been being set up for the last two to three decades, Ozette has been able to capitalize on.
I had the [00:04:00] opportunity to meet my co-founding team out of Raphael Gottardo’s lab, as you had mentioned, who was the Director for Translational Data Science Integrated Research here at the Fred Hutch. It was a whole new place where you’re integrating across data sets the science meeting computational biology. My two other Co-Founders Greg Finak and Evan Greene were all part of Raphael’s team, they were senior staff scientists at the Hutch as well, who joined as my CTO and VP of Data Science.
A lot of this actually spun out of this need to look at really high dimensional single cell data, as you had mentioned. I think the last few years we’ve seen advances in instrumentation that has been able to generate large volumes of data in a quick amount of time.
So I met my, I met my co-founding team and we sat down, as an opportunity to look at the work that they had pioneered in the field of single cell analysis. [00:05:00] Now, again, there’s a lot of background context here that I think is important to point out in the sense that, you know, biology had never had a large amount of data until just recently. Biologists, never had access to this large amount of data and the tools to analyze it. Now we’re creating such a large volume of data that some of our partners will tell you that they’re only looking at 10% of the data. In fact, they’re leaving 90% of the table and the fact that, Greg, Rafael, and Evan had pioneered a lot of the tooling to be able to do the analysis of single cell work allowed us to be in a unique position.
When we’re talking about single cell analysis, what we’re talking about is that immune system. Coming back to why the Hutch was a unique player in all this because of the resources around and uncoupling with the immune system has been able to do, we’re able to use our technology to really measure individual cells [00:06:00] as it’s coming off an instrument. That allowed us to not only just resolve what the immune system is doing.
As you know, I’m a physician by training and I still spend part of my time at the hospital taking care of hospitalized cancer patients. There’s this frustration of not knowing how patients are responding to treatment and understanding better how patients are optimized for the right treatment is super important in today’s environment. So the right treatment for the right patient, what does that mean? How do we find that? What are the instruments that are necessary? What are the platforms doing? So what we’ve essentially been able to do was, build a platform that’s automated the workflow for scientists in order to unlock the power of the immune system. That platform has also allowed us to use power of machine learning in order to unlock the power of the immune system.
[00:06:53] Chris Picardo: So, Ali, I think it’s really interesting, I’ll pick up on a thread that you just mentioned, which is that over the [00:07:00] last four years, five years, it doesn’t really matter, there’s been an explosion in the volume of data, right? The instrumentation for collecting and putting more kind of granular detail on this single cell data is really catalyzing that volume, how, without Ozette, would people analyze that data?
[00:07:23] Ali Ansary: They’re all analyzing it manually. This is the other part that gets us absolutely excited. The field of science is ripe for innovation to automate that workflow of data. So traditionally we have scientists, computational biologists, analyzing high volumes of data, but the dimensionality only recently has become very high. Before we’re looking at maybe only a handful of different parameters.
What I mean by that is, a computational challenge brings you back to your old math days. So two to the nth power, “n” being the number of proteins you’re measuring, or parameters, and two, whether it’s on or off, or there or not. [00:08:00] So in the past, we’re only measuring six or eight different components, so two to the eighth power means the search space is pretty feasible manually, but now that we’re measuring 30, 40 hundreds of different parameters, that search space is too large to do manually.
This is where the platform that we’ve developed allows you to do not only the analysis in real time of this high dimensional data, but allows you to explore that data and understand the visualization of that data and the impact of response rate to therapy. So,to bring it back, what you were saying is that there’s a large volume of data that’s being produced that was just being done manually.
It’s funny becauseI think that the bar for innovation is actually very low right now. That’s because we are only now starting to begin to see the opportunities that exist.
[00:08:56] Chris Picardo: Just put an example on the last thing you just [00:09:00] said, Ali, you know, give some color around the type of insight that you can derive if you can look at the data in higher dimension, versus when we used to only be able to look at four or five parameters.
[00:09:13] Ali Ansary: That’s right. So this is the part that gets me most excited as a physician because now I can determine response rate. I can now determine whether or not my patient is going to have Cytokine Release Syndrome or Tumor Lysis Syndrome, it’s that anticipation of treatments. When you look at these immunotherapies, these checkpoint inhibitors, for example, you’re unlocking the brakes of the immune system, right?
Now all of a sudden it’s going at a hundred miles an hour and, you know, Usain Bolt ultimately runs out of energy as well. With our platform you have, now, this ability to predict when the immune system is going to shut down and then you can augment that system as well.
Not to be too heavy in the analogies and I’ll have to give credit to my mentors for these analogies, it’s not something that I’ve come up with my own, but what we’re seeing is [00:10:00] that we can’t just be treating patients with single therapies any longer It’s going to be a combination of different therapies in order to treat an individual. It’s much like playing Beethoven’s fifth, it’s not just with one violin and you have to know when the various strings are coming in, in order to be able to have a beautiful symphony, and that’s what we’re seeing in immunotherapy today.
[00:10:22] Chris Picardo: Is it fair to say then Ali, that, Ozette sort of gives you that lens and that insight generating ability to be able to do some of the things you just said?
[00:10:34] Ali Ansary: Yes, so the opportunity here is not only in just automation of the data that’s coming off the instrument, but it’s also to learn from that data. It’s to be able to build a corpus of data against different disease processes, and then to be able to integrate across different data sets. You can use NLP to query against existing data against the publications, against publicly available data. There’s a large component of computer [00:11:00] vision with actual visualization of the data that’s being generated. The machine learning itself allows you to optimize that data generation. The insights that you’re now providing are just as valuable as the instrument that’s being produced on. Your Porsche is only so great as the driver behind it, otherwise it’s just a car that takes you from point A to point B. So, we’re positioned really uniquely to take advantage of single cell data that’s being generated in high volumes.
What we’re talking about here is at that protein level. We know that we are still developing therapies that target proteins, the FDA recognizes that it’s the proteomic data that’s the most valuable. You don’t necessarily submit transcript or you don’t make data for the FDA to approve, and a lot of the instruments that have really catalyzed the science, everything from genomics and transcriptomics and spatial omics, still are really early R&D discovery instruments, and [00:12:00] they will continue to increase in throughput. We’re seeing this, and that’s why the platform that we’re developing has a multi omics approach, but our primary focus has really been at the single cell proteomics level to really resolve each individual cell type and building these corpuses of data in order to be able to drive.
The one last part to not discredit is that, the instruments that measure the single cell proteome plays a role in every single part of the drug discovery pipeline from discovering and R&D through early and late phase clinical trials, all the way post FDA approval. These instruments have been around for the last 3-4 decades. So we really are primed to be in a place where there’s an accepted methodology and instrumentation for generating data, for us to be able to take advantage. There’s so much information and data for us to utilize and build upon that. We’re really primed uniquely right now [00:13:00] for this.
[00:13:03] Chris Picardo: One way, now that you mentioned, that I like to think about how to frame our areas of interest when we invest in intersections of innovation, is to think about sort of two broad buckets. One is, companies that have a method to generate totally novel data. A way that you wouldn’t have been able to see this before and then be able to apply machine learning or some sort of software to that.
The second is companies that are saying like, “hey, there is an explosion of data, but the tools and the ways that we’ve been looking at the data is not keeping up with the volume and where it’s going, and that we actually have to have a lot of insight generation mechanism to take advantage of what’s going on there”.
I think in some ways, those that sit in both of those buckets, but certainly there’s such an interesting thing to do in the second bucket where, like you said, there’s so much of this data and the lens has been manual, and now we’re switching it to machine [00:14:00] learning and insight generation.
So, given all of that, I thought it would be interesting to talk a little bit about the founding of Ozette. I think one thing that, that you guys did that’s fascinating and I just love your learnings on is, this company was formed in a trifecta process where Rafael, Evan, and Greg had been building some of the core software that underlies Ozette, at the Hutch for years. You were at AI to, trying to understand where are the interesting places that I can engage in this intersection and you guys came together and you built a company that has really deep academic roots.
I’d like to just have you add some color on how you think about building a company and bridging it out of the academic world and having to bring it into the more commercial use case and pharma or [00:15:00] customer facing world.
[00:15:01] Ali Ansary: I love that you asked that question, because in the life sciences, that academic affiliation, that academic validation is absolutely important. It’s a field where you cannot just sell vapeware, that your science has to be validated by your peers. You have to be able to replicate it and over validate it, based on publications in the science itself.
So to have an academic foundation has been crucial to what we’re doing. In fact, I think it has given us a strong platform to build, so we already had, I think, the back of the envelope was nearly $6 to $8 million in grant funding to be able to build the tooling that’s necessary. That gave us significant advantage versus anyone else who’s trying to build any kind [00:16:00] of competing platform to what we have.
It also allows us to actually be more resourceful too, because we know that with grant funding, academic institutions, you have to be nimble, you have to be fast, you have to be resourceful, because you only have only a finite amount of time and money to be able to build a technology.
The academic spin-outs have been advantageous to us. It allowed us to partner very quickly because of the reputation that Rafael, Greg, and Evan had built over the course of a decade. It allowed us to take advantage of an open source community that we’ve been continuing to support, and building that reputation saying, “hey, we’ll continue to support the open source, which is very critical to part of our mission [00:17:00] and business, but at the same time, we can build a significantly better version and automated version of what our open source tooling provides”.
We were positioned very uniquely, but I think in the life sciences as a whole, that academic affiliation is really important. It allows you to have a place of creativity, a place of research, a place of being able to build out, iterate quickly, experiment, and then bring alive. So, for us again like I had said earlier, I think it really positioned us in a unique place. I think in the field of life sciences you do need that those academic partners. Many of our early studies that we’ve been partnering on are through academic cancer centers, some are industry sponsored, some are not, but it allows people to become more aware of what you’re working on.
[00:17:54] Chris Picardo: How do you think about the team building when it comes to that? You’ve got a couple of core [00:18:00] team members who are at the company and you have a couple of core team members who will stay in academia and so how does that, as CEO, inform how you think about building a team around that kind of core academic group who really got the core product where it is today?
[00:18:23] Ali Ansary: Chris, I love that you asked that question. What we’ve been able to do at Ozette is position ourselves, not just as a life science company, but as a company that is bringing core values and missions that I think resonate really well with traditional engineers that have been at companies that are just making more money off advertisements and clicks. What you’re seeing is that engineers are tired of this and want to be able to have an opportunity to contribute [00:19:00] to something to a vision that’s beyond them.
At Ozette, it’s really creating an ecosystem and a dynamic where you have engineers, not necessarily executing a scientist’s idea or vision, rather creating an environment where the two can learn from each other. So you have these “aha” moments and you have this true curiosity and collaboration to be able to push a vision even forward. We’re lucky again, there’s a foundation technology that we’ve developed, but that’s version one. We’re planning for version two, version three, but the version 100 is going to come from these moments where you have talented engineers working across disciplines. This is where we have a lot of different opportunities to be able to educate whether it’s active learning or passive learning.
[00:19:49] Chris Picardo: I think that’s awesome. One thing I’m curious about is Ozette’s in a pretty specific subsection of life science, [00:20:00] right? Single cell data is its own world and field of study and we’ve seen that in a lot of ways. We’re bringing engineers with no background into something that’s really deep.
Just talk a little bit about the interaction between the engineers that you have on the team and the scientists and how they work with each other and what you sense is going on there.
[00:20:28] Ali Ansary: The interaction is something I have witnessed only a few times. What it is, is creating an environment of free flow of thoughts and questioning. There’s this culture of authenticity that we’ve been able to create at Ozette. I have no idea how it came about, but it’s been very unique in the sense that there is no such thing as a silly question.
It’s an [00:21:00] opportunity to truly bring in world’s experts who have developed everything from a front end to a backend with computational biologists to ask questions. As long as we’re asking questions, we start focusing on what are the right questions we need to ask and having that opportunity to have candor and challenge, and then come to a resolution quickly and then going and execute an idea. If it fails, you come back and iterate, understand why you failed in order to build forward.
Every component of any startup is risk and how do we mitigate risks? How do we create alignment? Being aligned that it’s okay to fail and coming back again to iterate on those processes. What we’ve also done is making sure we’re aligned on our vision and our vision is very much finding the right treatment for the right patient.
We’re in such a unique place to be able to help catalyze, [00:22:00] and it sounds scary, but the cure to cancer. It’s really a unique opportunity for someone to join and say, “yeah. I can get with that”. It sounds surreal for me to even say that, but the partners we’re working on are developing some of the most cutting edge therapies. so much talent that we’ve been able to bring together that it’d be a waste to not truly have a strong vision that we can try to execute here.
[00:22:31] Chris Picardo: Yeah, no, that’s awesome. I love to, that you mentioned “stupid questions”. I feel like we talk once a week and I asked you lots of stupid questions about science and it’s clarifying for me. Hopefully, sometimes it’s clarifying for you too. I do wonder that now that, teams are becoming really cross-functional in this space, especially if that’s a little bit liberating for both the engineers, the product people and the scientists where someone’s like, “hey, you spent a ton of time in [00:23:00] PhD, which is great. You know way more than anybody else, but how many people ask you the dumb question anymore?” I wonder, and this is probably more of an open question, but if we’ll see that being clarifying and freeing as you guys move down and continue to create what you’re building,
[00:23:19] Ali Ansary: I think it is, and more than ever, we have to just remain intentional about it. We have to be intentional about even a job description, so we’re not excluding a particular individual for them to say, “oh, I don’t know if that place is for me”. We have so much to grow and develop. If you were going to ask us if we needed a designer 18 months ago, I would have said, “yeah maybe”. But it was probably one of the best decisions we made because as soon as we put Calvin in the room, you have these aha moments that just come within the first 15-20 minutes, and everyone leaves aligned and [00:24:00] energized with who has been able to deliver really strong products in the past.
There’s a lot of skills and expertise that I think we tend to overlook that are not always in the life sciences. That’s, what’s been really important is how do you bring those in even more now than ever before?
[00:24:20] Chris Picardo: That Yeah, that brings me to my next question. We’ve got all these cross-functional team members. It’s been awesome and we’re building, what’s largely a software company, or at least on the UX side looks like a software company to a world that hadn’t had a lot of software companies that operate like this.
What are you finding challenging or liberating, or what are you learning about trying to build a software product with this type of data, in this world where the end user is at the end of the day, aren’t often people who have been sold or used 50 software [00:25:00] products every week?
[00:25:01] Ali Ansary: This is a good question. The reason it’s a good question is that what we’re learning is the end user and the decision maker about adopting a technology, are often different people. Sometimes you’ll have a scientist who will climb the ranks because they’re on maybe a scientist management track and they’ll grow into a director, senior director or VP, but when you become a senior executive at a big biopharmaceutical company, you want to know whether or not your billion dollar investment in a therapy is going to be successful in successfully to be able to treat patients with a disease. You don’t want to see that therapy fail because [00:26:00] it’s a heavy investment, but the steps needed to bring that therapy to market, the early R&D and discovery, and every component of those clinical early and late clinical trials have different users involved with that.
I think one of the biggest challenges is your end user is not oftentimes the person who you’re selling to. You have to find your internal champions who have these aha moments. We’ve had a handful of different times where we’re presenting to a team of 40 or 50 scientists in a room and it’s one person who says, “oh, I get this. This is incredible. This is so far ahead of where we’re at”, and I’ll get an email right after saying, “this is incredible”. I’ll say, “yeah, absolutely! This is why we all left our full-time jobs to do this”. end user is not always going to be that champion.
[00:27:00] Navigating industry is really important. Industry is siloed. There’s a lot of turnover. There’s a lot of old data that’s sitting on the table. There’s a lot of different competing priorities. And only now we’re beginning to see a lot of this data becoming digitized and that allows us to have a lot of freedom to explore through different avenues.
[00:27:26] Chris Picardo: I’m lucky. I get to spend a lot of time with Ozette every week so I’m very close to the product roadmap and what we’re thinking about on a daily basis, but I am curious when you think about it from a user facing perspective. You’re building software for, at the end of the day, for scientists and drug development and discovery teams who are trying to find new therapies or understand how patients respond to different treatments.
That’s a lot different than building software for managing your workflow or optimizing your ad spending or any of these other things. [00:28:00] So I wonder, when you think about, what are the implications for design and visualization facing them? What have you had to rethink, or is it really just a blank slate where you were saying, “hey, there’s a lot of stuff we can do here and our partners are so eager to work with this software that we can experiment and some new modalities”?
[00:28:21] Ali Ansary: I think that we’ve struggled a little bit about this. We’ve been building out this partner advisory committee where we have scientists helping to shape and give us feedback. I will say that the talent on our computational biology team or data science team has allowed us to really say, we’re actually building this for ourselves, which is a pretty high standard. As long as we know that we’re building a product that we want to use, we will see it widely adopted. There are different components that a scientist will want to see or do because scientists are creatures of [00:29:00] habit and they want to still be able to maybe export the data and interacted on a legacy system, which is great and we’ll create that for them. But at the end of the day, scientists want to be able to see their data, interact with it and the first thing, anytime a publication is going to be a RIN, is the visualizations. You have to have the graphs that are available to be able to then write the story around it.
That’s really what we want to be able to deliver, is that ability to understand that complex data. I think it alleviates a lot of the time that the scientists are spending to do the analysis of data. We go straight to insight generation and these insights help drive really important decisions that again, determines whether or not a therapy will come to market or not, because we can help find the right patient for the right treatment.
[00:29:54] Chris Picardo: I think it’s so cool when you think about the visualization side. [00:30:00] One way to talk about what Ozette has already achieved is moving from looking at graphs and drawing circles by hand on a piece of paper to generate an insight because you’re using machine learning to identify factors and dimensions that you physically could not see.
So you couldn’t have drawn a circle around it anyways, because you didn’t know it was there. I think it’s interesting. I think this optionality and the ways that that will end up being translated through software, it will be an interesting paradigm for us to help create.
I’ve got a couple of last questions to wrap this back to the beginning. I’ve talked in the beginning about the power of what you can do with single cell data and Ozette is really facilitating, generating tons of new insights out of that. If you think about, call it the short to midterm, not the super long-term, but the next three to five years, where do you see [00:31:00] this world of single cell analysis and immune profile, and is this powered by Ozette, going and what are the interesting things that you are super excited to see happen?
[00:31:12] Ali Ansary: Another way to probably ask me that same question, Chris is what keeps me up at night? In the short term, we make it very clear that we’re here for long-term partnerships. We’re here to help uncover the large amount of data that has been otherwise left on the table and to provide novel insights that we actually don’t even know how to value because you’re getting so much information.
In the next half decade we’re going to be at a place where we’re actively monitoring every individual on clinical trials with high dimensional data and that data is not only helping [00:32:00] with pharmacodynamics, pharmical kinetic information, it’s determining response rate. It’s ensuring that the right treatment is getting to the right patients and Ozette will be the default platform all the way through for any single cell data. We’ll continue to build in our transcript omics and in spatial omics because we know this is the way the field is moving towards. What makes our platform and machine learning very unique is that, it’s unsupervised and it allows us to continue to be agnostic of instrumentation.
As a physician, my biggest frustration has been looking at a complete blood count or a complete metabolic panel, spend only a couple of seconds looking at it, and then I look at my patient and say, “all right, they’re healthy enough to get treatment or not”. We’re going to fundamentally change that because the insights generated from someone’s immune system is going to allow us to [00:33:00] determine that response rate in real time. We won’t need to wait six to eight weeks any longer to determine if you need a pet CT scan, to determine response rate. We won’t need to go back and rebiopsy a solid tumor because we have these immune markers that we’re able to extract. That’s in the short term and I know the short term is right around the corner, but in five years, we’ll have a lot of opportunity to mature with a lot of therapies in the field of immuno oncology.
[00:33:34] Chris Picardo: That’s a good vision. I think that if we accomplish that in this short term, that would be pretty incredible and not just for Ozette, but for patients and I think that’s, at the end of the day, what this is all about. That feels like a really good place to end because I’m not sure you could say it any better and so Ali I want to thank you a ton for coming on the podcast.
[00:33:55] Ali Ansary: you, Chris. I appreciate it. I always have fun doing these.
[00:33:57] Chris Picardo: We Yeah, do this and we [00:34:00] just don’t record it on a weekly basis so it’s fun to put it down and make something for other people to listen to, but really appreciate you taking the time Ali, and thanks for joining us.
Thanks for joining us for founded and funded. I’m Erika Shaffer from madrona venture group and i’ll be talking to you next time
Related Insights
