https://youtu.be/-7x5kBxJ5ig
Ben Larman, Scientific Founder and Chief Scientific Officer (CSO) of Infinity Bio, is driven by a passion for advancing human health by decoding the complexities of the immune system.
We discuss Ben’s Antibody Reactomics Framework and his Complex Data Delivery Framework. The Antibody Reactomics Framework leverages DNA barcoding to analyze immune responses, providing researchers with groundbreaking insights into human health and disease mechanisms. The Complex Data Delivery Framework streamlines the process of communicating scientific findings by focusing on understanding customer needs, analyzing data, presenting positive results, and combining findings to deliver customized insights. He also shares how AI is transforming biomedical research and the importance of refining a biotech company’s narrative to effectively engage diverse audiences.
---
Simplify Complexity with Ben Larman
Good day, dear listener, Steve Breda here with the Management Blueprint podcast. And my guest today is Ben Larman, the Scientific Founder and CSO of Infinity Bio, a technology company that measures individual immune responses against all known human viruses, autoimmune conditions and allergies. Ben, welcome to the show.
Thanks, Steve. Great to be here.
My favorite question for you. So what is your personal “Why” and what are you doing to manifest it in your company?
Great. Well, yeah, I'll start by saying I guess I became an immunologist by training because I believe there is an incredible opportunity to improve human health by better understanding immune responses and the immune system. The immune system is incredibly complex. It's sort of, we're just scratching the surface now in terms of our understanding and how it's connected to so many different health conditions ranging, as you mentioned, from infections to cancer, to autoimmune disease, to allergies. The ability to move technology forward in a way that allows us to take better measurements and understand human immune responses at a population scale, that's always motivated me. And so that's what we formed Infinity Bio to accomplish by being able to read antibodies.
Okay. And when you read antibodies, so how does that inform you? How does it help you fight immune conditions? I guess you can call it.
Yeah, so antibodies, I think we all sort of have seen the cartoons of these Y-shaped molecules. I think what a lot of people don't appreciate is the massive abundance and diversity of these molecules in our blood at all times. So there’s roughly 100 trillion antibody molecules in every drop of blood, and they really store the information from all of our prior immune responses, going back to your vaccinations in childhood. And one way to think about them, they're expressed by cells in our immune system in response to very specific targets. They have these ridges and bumps on them, and that's what allows them to very specifically recognize their target, whether it's flu or a cancer antigen. And so what we've done is to create a technology that you can think of like a record player where the record, your old vinyl records with those grooves and bumps read out by a needle, that's sort of what we're doing with molecules, reading out those bumps and grooves to know what they bind to and get a picture of your immune system. Antibodies are really a wonderful window into your immune system and everything that it has tried to accomplish over your lifetime.
So that's an amazing complexity. You said trillions of antibodies in a drop of blood.
Yeah.
That's very, very minuscule and very numerous. So how do you measure this data or process even this data? How do you then communicate this data is about?
And actually you develop a framework around this. So can you tell me a little bit about how that came about and what's its significance is and how does it work?
Yeah, so the approach that we've taken and that I've sort of worked on since grad school is to try and convert the antibody target interaction question into a DNA synthesis and sequencing problem, which is something that is much more approachable because DNA technologies have advanced so dramatically over the past couple of decades, both in our ability to create DNA molecules and our ability to read them. And so without going into too much technical detail, we basically have created a system, a technology, for producing large collections of proteins that could be targeted by antibodies and putting DNA barcodes on them. So, these are just sequences that associate what that target is. And so we can take someone's blood sample and we basically mix it with this collection of molecules and we pull out the molecules that are bound to the antibodies and use DNA sequencing to read it out. It allows us to take advantage of those advances in DNA technologies to measure all the different things that someone's antibodies bind to. And so, just to give you a sense for the scope of a particular test that we offer at Infinity Bio for all the viruses, as you mentioned, we test antibodies against about 285,000 different viral targets. These are proteins, fragments of proteins that are associated with all the viruses known to infect humans. And so, for each sample, we get a result that says, okay, these are the proteins that are recognized by that sample's antibodies, and these are the proteins that are not recognized by that person's antibodies. So, at the basic level, we're generating sort of like a report for each sample. We typically work with investigators who have sets of samples, some from individuals with a particular disease, for instance, like type 1 diabetes, and a set of samples from individuals that don't have that. And they would like to know what is the difference between these groups of individuals in terms of which viruses they've been infected with, and maybe to an even more detailed level, during those infections, what were the specific targets of those antibodies? And by comparing the groups, we could identify things that may actually be mechanistically causally associated with the ultimate disease that these individuals developed. And so communicating that data is a challenge. It's a very deep data set that we typically will deliver in a way that can be opened with Excel, but they can be very large files depending on the number of samples that we've tested. So we have an amazing informatics team at Infinity Bio, but it's really been something of a learning process for us over the last year as we've been operational and returning data sets to folks is how can we do it in a way that provides them the most insight into the question that they're really trying to ask. And I think we've learned several lessons that we'll be able to improve the way that we deliver data and insights to our customers going forward.
I think that AI will be able to provide a really important resource in this area.Share on X
Because really what people want to know a lot of times is what is the difference between the groups and then how is that related to all the scientific research about whatever that difference is? So, it may be the case that we identify three different viruses that are associated with the disease. What's known about that already? It may take someone a long time to go in and mine the literature to figure out what is that connection, whereas an AI, generative AI, can help dig out the papers that are most meaningful, most cited, or whatever it is, to help them really interpret that information.
So, Ben, what are the steps then of your process to communicate this really complex information that could potentially overwhelm the recipient?
Yeah, great question. So, it's a process that will change because up till now, we've really been returning the datasets without as much conversation upfront as we should have been having to really understand what they want to learn from their data. And going forward,
we'll want to engage our customers in a more thorough way and prepare an analysis in that first data return.Share on X
Right now, we're returning data, we've been returning data in a way that's sort of generic. And we say, okay, take a look at your data. Let's get together and figure out how to answer your question. Going forward we'll put more of an emphasis on let's come up with an analytic plan ahead of time so that when we return those results in the first instance, you get your generic reports and your generic data, but we also give you an analysis that's specific to your question.
Now, one of the things you mentioned in our earlier discussion is that when you present a mix of positive and negative results, then customers often get confused and sometimes less is more. The more you give them, the more confused they are and you have to somehow curate the information so that they can actually use it. How does that work?
Yeah, we've all had the experience of going to the grocery store and looking for butter and seeing 13 different options and being overwhelmed. And that has definitely been the case for some of our customers returning multiple sets of files that they're sort of not sure which ones to go into first if they want to be very thorough. And we need to do a better job of distinguishing for them. Here's the key results. And in antibody reactome profiling, we take these very large libraries, these collections of proteins, most of the data is actually negative. And so these big files that we're giving to people, it's mostly negative, but what they really care about is the positive data. And we'll be doing a better job of integrating that positive data across the different service offerings because sometimes people will combine services for certain projects. So we'll just be doing a better job of distilling down the positive data that provides the most insight and separating out, making it available to them,
