AI-Powered Biology? Dr. Shana Kelley, President of Bioengineering & Head of Biohub, Chicago
What does it mean to be all in on AI powered biology? Chan Zuckerberg Initiative's Doctor. Shana Kelley is my guest. She tells us about what they're working on and what that means for the future. Doctor Kelly, welcome to the program.
Dr. Shana Kelley:Wonderful to be here.
Dr. Moira Gunn:Now I keep hearing that the Chan Zuckerberg Initiative is all in on AI powered biology, and I think we all understand what biology is. What's AI powered biology?
Dr. Shana Kelley:Yeah. Well, biology, as I'm sure you're aware, is a discipline where we form hypotheses about what might be going on. We do experiments. We interpret the experiments, and then we try to come up with holistic views of what's happening in cells, what's happening in the human body. And that process, it turns out, is very labor intensive and a little slow.
Dr. Shana Kelley:Biological discoveries take many, many years. There's a lot of different moving parts that have to come together. And our approach now is to use AI to really accelerate and to take very large data sets that will go above and beyond what we can do either with human interpretation or even bioinformatic interpretation. And we're gonna use machine learning and artificial intelligence to really speed the process of understanding large data sets and being able to simulate biology so that rather than sending armies of people into the lab, we can mine models and come away with those insights much faster.
Dr. Moira Gunn:But it has to start at some level in the lab because you've got to deal with tissues and cells and the like, right?
Dr. Shana Kelley:Yeah, the lab is not going anywhere. So we need to generate large data sets in the lab to train the models. And then as we get predictions that are made by our models, we have to go back to the lab and we have to validate them and make sure that whatever the model is pointing us towards is real. And then if we have an insight that let's say is going to help us develop a new therapy or a new diagnostic, well then we got to get in the lab and really bring that to fruition. We're not going to have these things just jump out of the models for us.
Dr. Shana Kelley:We're going to have to do the work to bring them to patients.
Dr. Moira Gunn:So let's go down then to the next level and that's where we get the all in part on AI powered biology. That's the Chan Zuckerberg Initiative. What is it? Who is it? And and what's it all about?
Dr. Shana Kelley:So ten years ago, Priscilla Chan and Mark Zuckerberg founded the Chan Zuckerberg Initiative, CZI, to really make an impact that would change our trajectory as humans and to really make the world a better place. Over time, they have been very generous in terms of the resources that they have put into that philanthropy. And the philanthropy did a variety of different things. Science was one of them. Over time, they decided to focus on science.
Dr. Shana Kelley:That's where they thought there could be the greatest impact and where there was a great need for the resources that they could bring to bear. And so over time, CZI has become more and more science focused, and now we're getting even more focused. And we are really concentrating on AI powered biology. And so we are all in on this. We are taking all the resources.
Dr. Shana Kelley:We're doing frontier AI, but we're combining that with frontier biology, right? The cutting edge of biology, cutting edge technologies in biology. And what we hope will emerge from that is frontier medicine and really paradigm shifts in terms of things that we can offer to patients or to the human population to keep them well.
Dr. Moira Gunn:Now I see that their big goal here is to support the science and technology to make it possible to cure, prevent, and manage all diseases by the end of this century. Now personally, I need it to happen faster. Right? Okay. End of this century.
Dr. Moira Gunn:I have never I have never seen a goal or a quest like this that was so inclusive and had a science and technology focus that just might do that.
Dr. Shana Kelley:I could not agree with you more. It's very exciting. And one of the things that I love about being part of this organization is that we're just incredibly ambitious. Like there is just unbridled ambition here. And it sounds challenging, right, to tackle all of disease, to make a dent in our high burden diseases and really change the trajectory, you know, of of our health as a population.
Dr. Shana Kelley:But that is what we are striving for. And we really only pursue things that have that transformative potential. We don't want to do incremental things. We want to take big leaps. And that is really the philosophy behind the way we work at CZI and Biohub, and it's incredibly inspiring.
Dr. Shana Kelley:It's wonderful as a scientist to be able to work that way.
Dr. Moira Gunn:Now this effort is organized around four core scientific grand challenges. Let's go there. Let's talk about them.
Dr. Shana Kelley:Sure. So we've identified four areas that we think are just excellent fields for us to focus our efforts, where there is white space, where there's an impact to be made using our approach, being very technology focused, AI focused, wanting to make breakthroughs in biology. And so we have one challenge that is focused on developing new imaging technologies that are allowing us to watch biology in action at an unprecedented level of detail. We have another challenge that is focused on reprogramming immune cells so that we can use them to really surveil the human body and fix problems within the body. And then we have another one that is focused This is where my work is based.
Dr. Shana Kelley:We're developing new tools that allow us to watch inflammation in real time in the human body, in tissue models, in organoid based systems, just to allow us to really watch inflammation and decode it learn everything that we can about it so that we can figure out how to reverse it in the human body when that makes sense. And then finally, we have a challenge related to creating unified models of the cell, virtual cells where we can ingest very large amounts of biological information, be able to do virtual experiments, and again, accelerate biology. So we have these four main areas, again, all frontier areas, and we're approaching these problems in a really different way, a completely multidisciplinary way where we bring people together from across scientific disciplines, across engineering, across computer science. And it's really exciting to be focused on these areas where there's a lot to learn and there's just so much impact to be had if we are successful.
Dr. Moira Gunn:Now we understand, I think, in general, but you mentioned the word organoid. What are organoids?
Dr. Shana Kelley:Doctor. Yeah. So organoids are a type of tissue model. So there are some biological processes that you really don't want to study just in a cell. Inflammation is a perfect example.
Dr. Shana Kelley:Inflammation happens when you have immune cells talking to the cells within human tissues and that communication is not going so well. And there's things being said that are not so good for the cells. And we can't just take single cells. We can't take cells in a petri dish and recreate that. We need a tissue model and an organoid is basically it's kind of like a mini model of an organ.
Dr. Shana Kelley:And so we bring different cell types together. For example, we can make organoids of skin. And so we take skin cells and we layer them up and we get exactly the same structure that you see in human skin. You get an epidermis, a dermis, and then you can start to do experiments there where you're in that complete tissue context rather than in cells in a petri dish.
Dr. Moira Gunn:So while in biology and in biotech, you would actually take skin and do these experiments. You'll do them in the computer, if you will, in computational and data form. And then you'll go back to the lab to say, did we get that right? And when you understand you got it right, then you can use the computer again and again and again for many different kinds of experiments.
Dr. Shana Kelley:Exactly. It's a feedback loop. It's a virtuous cycle, right, of computation and work being done in the lab. We try to do as much as we can with computation, but ultimately, again, to validate and also to bring things to life. You know this, that to develop a new therapy, there's a lot of stuff that has to happen in the lab to validate that this is something that we should take towards patients.
Dr. Shana Kelley:And we'll continue to do that, but we're already seeing that we can accelerate all of the kind of random walks that we do in the lab or the random walks of hypothesis generation. Right? It takes a lot of wrong turns very often in experimental research to get from point a to point b. But with computation, you can go on a much more direct route because you're sampling all those right turns and left turns and things that you probably shouldn't bother to look at and then get to that right conclusion.
Dr. Moira Gunn:And your computation could go on in the cloud, you know, while you're sleeping. Doesn't get
Dr. Shana Kelley:twenty four seven.
Dr. Moira Gunn:Yeah. Exactly. Exactly. Now I want people to understand that the AI you are talking about is more than the LLMs, the large language models, which, you know, many of us use every day either to generate text or when we, you know, perhaps Google something, you know, say, hey. This is what this says.
Dr. Moira Gunn:All of a sudden, we get the the AI input on that, or we'll have at the top of our email, you might see this email thread will tell you that this person talked to this person and this person did this. So those are part of the large language models. They are looking at language, analyzing it, generating things from there. And that is only one part of AI. And yet you are developing new LLMs in support of what you're doing.
Dr. Moira Gunn:So let's expand a little on that AI.
Dr. Shana Kelley:Absolutely. So LLMs have already been shown to be very powerful in science and for biological research because there are languages in biology. There's the language of protein sequence. And you can use LLM like approaches to predict sequences or to predict structures from those sequences. And that is something that is progressing, I think, extremely well.
Dr. Shana Kelley:There's other areas where people have analyzed transcriptomic data, other types of omic data that way. But we do want to look beyond LLMs. Right? Language is pretty linear. Right?
Dr. Shana Kelley:You're just looking in one dimension for what word comes after another one, same thing. You could look at some biological processes that way. But biology is multidimensional, like really, really multidimensional, right? There's so many things going on at one time. There's so many different ways that biology reads things out and and the dynamics are not linear at all.
Dr. Shana Kelley:And so we're gonna have to use different approaches to be able to capture that if we wanna do it accurately. And so we are certainly looking at using anything that already exists. We can certainly put that to work. But we are fortunate at Biohub to have incredible computer science expertise that I think if there are new types of AI that need to be brought in the mix to solve these problems and help us accelerate forward, we will be able to do that because those are exactly the kinds of people that we have working here.
Dr. Moira Gunn:Now we have four grand challenges here, but CZI has three Biohubs. Now how does that work?
Dr. Shana Kelley:Great
Dr. Moira Gunn:question. Did somebody get two? How did that happen?
Dr. Shana Kelley:Great question. So the grand challenges did arise out of these geographical Biohubs that were put in place over the last ten years. But actually, we just reorganized a bit, and we are now all one organization called Biohub. And we work seamlessly. We work across the country, and we work on these grand challenges in a way that is not defined by geography.
Dr. Shana Kelley:It's really defined by who has the right solution, who has somebody that can plug in exactly the right expertise into this problem. And I think that's one of the things that's really exciting about us as an organization is that we have hundreds of world class scientists, scientists, engineers, again, like computer science types of folks, and we're just all there working on these incredibly ambitious problems. And so, you know, this is an unprecedented structure, I think, for a biomedical research organization and and institute. And I think we're gonna be able to produce really new insights because of the multidisciplinary, because we have people coming from so many different perspectives and and with the different backgrounds being put to use. So it's it's an incredibly exciting model.
Dr. Moira Gunn:At the same time, we all know everybody doesn't have to be colocated all the time to do creative things, but you have to be able to have these hubs in places which can draw from a lot of talent. You can't put it someplace that's out in the middle of nowhere. It's not a data center that you you just you get out there and you plug it in, and there's a very limited staff that's needed to run it. So where are the Biohubs, and why are they there?
Dr. Shana Kelley:Great question. So the Biohubs got started basically in three locations in the Bay Area, so around San Francisco, in Chicago, and in New York City. And when each of the original Biohubs were started, they were done so in partnership with great universities. So in San Francisco, there was a partnership between CZI and UCSF, Berkeley, and Stanford. In Chicago, we have the University of Chicago.
Dr. Shana Kelley:We have Northwestern University, and we have the University of Illinois Urbana Champaign. In New York, we have Rockefeller, we have Columbia, and we have Yale. And so we have these partnerships, and we also have the incredible talent pool that is associated with universities that function at that level and that have the level of research activity that these universities have. And so this is part of the magic of the model. You're right.
Dr. Shana Kelley:We're not just out in the middle of nowhere just starting something from scratch. We are leveraging world class universities so that we have the talent, we have the back and forth with the great faculty that are at these universities. And it's, I think, part of what's really special about the model and what makes it work exceptionally well.
Dr. Moira Gunn:Well, you are the president, as we've said earlier, of bioengineering at Biohub Chicago. But and right now, you are in Redwood City at the headquarters, if you will, which everybody's getting together this week at the time of this interview. But you are also a professor at Northwestern, and I might add across three departments, chemistry, biomedical engineering, and biochemistry and molecular genetics, which speaks to the breadth of the kinds of things you're doing. And there may be some people who are saying, who is this woman? How could I possibly understand anything she's talking about?
Dr. Moira Gunn:Well, you've already understood everything she's already said. So, let's go to specifically what you are doing at Biohub Chicago. Tell us some of the interesting projects you're working on there.
Dr. Shana Kelley:Sure. So in Chicago, we think
Dr. Moira Gunn:a
Dr. Shana Kelley:lot about inflammation within the human body, and we develop new ways to study inflammation. It's a complicated problem to kind of take apart and to dissect. There's the immune cells that are doing their things, sometimes causing problems. There's all the tissues within the human body that interact with immune cells, and it's that interaction that can produce inflammation. And to study this, you need to be able to look in real time because this is a dynamic process.
Dr. Shana Kelley:You can't just take a snapshot and be able to understand it. You have to have a way of looking at intact tissue because that's what the immune system is interacting with. And then you also really have to have some spatial information to kind of understand how inflammation kind of spreads and how you might have inflammation in this part of the body or not that part of the body. And so in terms of technologies that can help capture this and can capture the molecular landscape of inflammation, we do not have a lot of tools. And so that is where we have been focusing.
Dr. Shana Kelley:That's what our work has really honed in on. And so we have two families of technologies that we're working on. One technology is amenable to the development of implantable sensors. So think continuous glucose monitor, right? Patch on the arm, little micro needle under the skin, and then you can measure glucose, right, in real time, all day, all night.
Dr. Shana Kelley:We're developing tools and already have really exciting proof of concept for an implantable sensor that instead of looking at glucose, it tracks inflammatory markers and can watch them trend up and down. We can do a handful of markers at the moment. So that's a technology that we're building on. We're also developing probes that, for example, could go in the brain to track neuroinflammation. There's really so many exciting directions we we can go with that type of approach.
Dr. Shana Kelley:But we look at small handfuls of of markers at a time. Five to 10 is probably the sweet spot. If we want to go and actually characterize what every single protein in the human body is doing during inflammation, we need different types of approaches that can really give us that coverage of the entire proteome. And so we have platforms that we refer to as live tissue omic platforms that allow us to look at the proteome or the metabolome. And the proteome and the metabolome are?
Dr. Shana Kelley:The proteome is the collection, thank you, of all of the proteins that do the work within the human cell and within the human body. The metabolome is the collection of all of the small molecules that are doing things in the cell. And so there, the oh, I'm talking thousands of different things. If you kind of go to the next level of detail, it's tens of thousands of different forms of proteins or small molecules. And so there, you need to go very broad.
Dr. Shana Kelley:You need to go very deep. And so that's another suite of technologies that we're developing.
Dr. Moira Gunn:So at one level, you're working on sensors to really check out what's going on in your body with that data. And you also have in the computation space models of what should be going on. So you should be able to see the differential. You should be able to see, for example, inflammation and the state of inflammation, which plays into so many diseases that we have today.
Dr. Shana Kelley:Exactly. And I think you've put your finger right on it that we develop new technology platforms that have the capacity to generate really large data sets, areas of biology where we have not had data previously. And then that really takes us to this place where we are very well equipped to start developing models based on that data and then mining them to see what new insights can be generated. But it's a loop. Then we can go back to the lab, back to these technology platforms to see whether we've come to a conclusion that's correct, and then we can keep generating data, keep advancing the models, hopefully accelerate the insights and the power of the insights.
Dr. Shana Kelley:And it's a loop. And I think our ability to work that way is what will allow us to make breakthroughs.
Dr. Moira Gunn:The idea of accelerating science, which accelerates diagnostics, prevention, treatment is really exciting. It's like, gee, does that mean there's gonna be a lot of new drugs? It's like maybe the same number of new drugs. It's just they're that much better. But I think it's very important to understand you're not doing that to make the world all Chan Zuckerberg drugs.
Dr. Moira Gunn:Chan Zuckerberg diagnostics, are you?
Dr. Shana Kelley:Not at all. And we are really not looking for the incremental, you know, for the let's give a cancer patient three to six months of additional life. Right? We want to make the big leaps. We want new paradigms.
Dr. Shana Kelley:We want completely new ways to treat disease and really cure disease is what we're striving towards. But even better, let's use all of these insights that we're generating to prevent disease. Let's get in there at a very early stage of things, maybe when inflammation is first starting to spark within the human body, and let's do something about that instead of treating end stage cardiovascular disease. And so that's really where we want to end up, is just in a place where we can cure and prevent and dramatically lower the burden of disease across the human population. Another thing that's really important about the way we work is that we want to disseminate everything that we're discovering, all of the insights, all of the data sets, we wanna get those out into the community so that other people can use them as well.
Dr. Shana Kelley:And CZI has a long history of doing this. They have incredible databases of single cell data that they have made available to the community. And that is the spirit in which we will continue to work. Anytime we can just push something directly out into the community, we will absolutely do that. We try to get all of our work preprinted so that people can see it even before it's published.
Dr. Shana Kelley:We make datasets available. And so this is not work that we wanna keep inside or keep to ourselves. We really want to get it out into the community to accelerate the progress that can be made using it.
Dr. Moira Gunn:Now I noticed your PhD is in chemistry, and over your career, you've expanded many times the disciplines that that you could work in and have contributed to. And I I look through a number of the people across the board. You've got PhDs across the sciences, and I've become more and more fascinated by the fact that we can describe our biology, the biology of our bodies, beyond the language of biology. We can use the language of chemistry. We can similarly, we can use the language of physics.
Dr. Moira Gunn:We can use the language of computer science. It's it's as if all of these things, all of these sciences now are coming together, and and that is new to me.
Dr. Shana Kelley:It's really exciting. I mean, it's it's convergence. Right? We have the kind of classical disciplines, chemistry, biology, physics. We've got engineering over here.
Dr. Shana Kelley:And the lines have been blurring for a while now. I, throughout my career, have always really wanted to work kind of like right at the interfaces. I think that's a very fertile place to explore and to find new ways of of putting things together. That's certainly how we work at at Biohub. And it also allows you to approach problems kind of almost from a naive perspective.
Dr. Shana Kelley:You know, you can look at a really complex problem, but you don't get bogged down. You know, you haven't been working on it for thirty years, and you don't get bogged down, like, with all the the baggage that can come with that. And you then get fresh sets of eyes looking at at these problems and and then just coming up with completely new ways to address them. And I think that is incredibly powerful.
Dr. Moira Gunn:Well, doctor Kelly, I wanna thank you for coming in. And, please know you're always welcome on Tech Nation. Please come back and see us again.
Dr. Shana Kelley:Thank you so much, Moira.
Dr. Moira Gunn:Doctor Shana Kelley is the president of bioengineering and the Head of Biohub Chicago for the Chan Zuckerberg Initiative. More information is available at czi.com. That's the initials czi.com.
