There’s an old saying, “If you want to catch fish, fish where the fish are.” Charlotte Deane, chief scientist of biologics AI at Exscientia, is leading the charge to redefine where the traditional fishing hole of biologics discovery lies.

UK-based Exscientia applies artificial intelligence to the process of finding small molecules and inked one of the largest AI partnership deals with a Big Pharma to date earlier this year. Around the same time, the company tapped Deane to redefine biologics discovery and design into one combined process.

“Everyone starts by what I would call ‘going fishing,’” she said. “You might go fishing in an animal model … or you might go fishing in a generated library. You fish and then you discover things that will bind. But what if we reimagined the process: Do you need to go fishing at the beginning? Could you design the actual biologic you want to make? And there are many, many advantages in doing that. An obvious one is fewer experiments, so you may well be able to go faster.”

Driving this transformation is AI technology, which allows for more sophisticated data gathering and algorithms for advanced problem solving. Deane said one way to think of AI is as a tool with the ability to generate novel molecules and antibodies that have never been seen before and uncovering the characteristics they have.

“It’s a powerful set of techniques that allows you to use data and extrapolate the best possible answers in terms of what experiments you should do next or what would actually be functional,” she said.

Deane added that while Exscientia already has had success in bringing a drug to clinical trials using its unique AI approach, it’s still early days in terms of biologics.

“At the moment we have prototype modules … to build better biologics and better antibodies,” she said.

Undaunted by the prospect of developing a new process to design biologics rather than discovering them, Deane says she is excited to translate that idea into reality and create a path for others from both industry and academia to follow.

In this episode of the WoW podcast, Deane shared how her days in academia gave her the confidence to pursue big scientific theories, what it meant to be awarded the Queen’s Honor for her work in the U.K.’s response to COVID-19 and how the experience led to her current role at Exscientia.

Welcome to WoW, the Woman of the Week podcast by PharmaVoice powered by Industry Dive.

In this episode, Taren Grom, Editor and Chief Emeritus at PharmaVoice, meets with Charlotte Deane, Chief Scientist of Biologics AI at Exscientia.

Taren: Charlotte, welcome to the WoW podcast program.

Charlotte: Thank you for inviting me. I’m looking forward to getting to speak to you about some of what I do.

Taren: I’m excited too. I’m really very curious about what a chief scientist of biologics AI role is. What is that role?

Charlotte: I have to be honest here, it really did take us a while to settle on the title for my job. The actual role in my head is really straightforward. It was to take Exscientia’s mission of what they had been doing for a while in terms of revolutionizing the drug discovery process for small molecules; how you automate and encode that entire process; and just do that for biologics. I sort of jokingly said that what we meant by that was recreate Exscientia all over again for biologics, but in some ways it’s exactly that. My role is to really build that team of people, the tech team, the experimental team, the kind of combined set of people needed to change us from the way we do discovery currently. People talk about biologics discovery to biologics design and really take the whole kind of process to one which is a full design process.

Taren: So tell me about this revolution. So we want to go from what we’ve done traditionally from drug discovery and recreate the process to be more efficient for biologics design?

Charlotte: Yeah. I think the easiest way in terms of thinking about biologics is that nobody in the world who does biologics, everyone starts by what I would call “going fishing.” What you do is in order to find your initial biologic to hit a particular target you’re interested in, so to hit your target molecule, is you use that molecule to go fishing with. You might go fishing in an animal model, so you might inject an animal with it and collect the antibodies that are raised against it; or you might go fishing in a generated library. So you fish; so you discover things that will bind. And then from then on, you may have processes which involve design and optimization which involve a computer. But reimagining the process from the point of view: do you need to go fishing at the beginning? Could you design the actual biologic you want to make? And there are many, many advantages in doing that, an obvious one is fewer experiments so you may well be able to go faster.

But another one is of course it allows you to target specifically where you want to hit on a given protein. If you go fishing with a protein, you only get hit where they happen to be but you might not want any of those; you might want it to be on another part of the surface. And it also allows you to control many of the other features when you’re doing this. So if you go fishing in a library, you cannot be certain that that antibody will have all the other features you might want it to have. So the idea here is really to design and build it in from the beginning every property you would like to have of the thing you’re trying to build.

Taren: And tell me about the artificial intelligence piece of this.

Charlotte: I kind of always think of this I guess because I come from a statistics background. Artificial intelligence is a tool to help us do all of this. What really has happened over the last few years, if we have seen the power of the algorithms that can be written in AI if you like problems of this type; and what I mean by that is problems where I can collect a reasonable amount of data, sometimes very large amounts, sometimes small amounts, and I can write down what I want to know or extrapolate to understand. So the easy examples in artificial intelligence where they came to public prominence were around kind of being able to recognize faces in images. But you can think of similar techniques in terms of recognizing whether this is a good antibody or not, if I have trained it on enough data, or you can think of it of being able to do techniques; and these are other things people have seen in images where you can morph one image into another using AI.

So you can also think of AI as a way of generating novel molecules you’ve never seen before, so novel antibodies, and seeing what shapes and characteristics they would have. So it’s a really kind of powerful set of techniques that allow you to use this data and if you like to extrapolate from that data or interpolate within it to give you the best possible answers in terms of what experiments you should do next or what would actually be functional that you are interested in.

Taren: It’s fascinating. There’s a big debate between the advantages of big data and small data. How do you look at that?

Charlotte: I guess I don’t really see it as a kind of debate between the two. In many ways, kind of the amount of data you have is actually controlled by how easy it is to do the experiment you’re interested in. In some places it’s really easy to collect lots of data; in other places it’s much harder. And to me thinking in a computational way, all that really does is change the types of computational techniques you might want to use. So there are computational techniques which are incredibly good at using very, very large data sets. There’s AI techniques for using these incredibly large data sets and extracting information from them, even if that data is quite noisy. If you have very small data sets, they tend to be less noisy and then you use very different types of techniques that allow you to extract information and knowledge from those smaller data sets.

So really in one sense it’s always true, the more data you have it’s like an increasing amount of knowledge. But the reality is the challenge is not always to generate more data, it’s to generate…I kind of like the phrase “machine learning grade data.” So data that you understand how it was produced, you understand exactly if you like what its errors might be so that you can encode and use it well when you want to make predictions. It’s really allowing the machines the data they need to learn – thinking of it that way around rather than having a preference for a particular data type.

Taren: Wonderful. And have you had some success so far in terms of using the AI in revolutionizing drug discovery? Have there been some positive outputs?

Charlotte: So, from Exscientia as a whole, there have been rather a lot of positive outputs. They have some drugs that have made it through to the stages of clinical trials and having the first AI design drug that went to clinical trials. In terms of the biologics, we’re still at very early stages here so I would say what we have at the moment are kind of prototype modules that we’re building to be able to build better biologics and better antibodies within all of this.

Taren: You also were very involved in the UK’s response to COVID. In fact, you were recognized with the Queen’s honor for your work. Can you please share what your role in working toward pandemic solutions was and what this recognition meant to you, especially given the Queen’s recent passing?

Charlotte: For me, like for many people, the pandemic was really kind of a weird and difficult part, difficult time. Back in 2019 I had agreed to be partially seconded into UKRI which is the UK’s research and innovation setup which provides basically all of the public funding to universities for research, but it also does public funding to industry as well. And I was there to help with particularly the engineering and physical science council. And then in March 2020, a date that I think many people will recognize as a point in time where certainly in the UK it flipped from a kind of ‘there’s something a bit odd happening in the world’ to the world completely changed for all of us because that was when the first national lockdown happened.

I was asked by Sir Mark Walport, who is the head of UKRI at the time, to work out how UKRI could mobilize its funding capabilities and by that mobilize the academic and industrial community across the UK to fund projects directed to kind of build our understanding against COVID and also all sorts of other bits around it. There’s both kind of directed research that you knew needed to be done, so part of that was how do we get money early into the vaccine projects and how do we get money early into setting up sequencing, but also how do we solicit ideas across the amazing science base of the UK at both industrial and academic and fund those very, very fast. Anyone who’s been involved in applying for research funding, and I believe this is true in most parts of the world, takes a long time to get your publicly funded money. You put in a proposal, it’s reviewed, it goes through rounds of review, it comes back and then eventually it gets funded and that might be 6 months to 12 months after you put it in; and in COVID that was not an option.

So within two days, well strictly speaking within four days but two of those were the weekend, we got up a website which explained how people could make applications to us with a new type of application form, an entirely new type of process to do this. And then the process was really difficult because what I had to do was I had to get to know the science that would be useful and that involved spending a lot of time talking to the chief scientific advisors to the various departments across government so being part of understanding what government needed to know in terms of helping this.

I was part of SAGE, the kind of advisory committee to government in terms of COVID as well, partly to be able to listen in but also because the areas I worked on are similar and useful within this. And then trying to get the community to put their ideas in. Our initial drumbeat was trying to be able to say to people within two weeks whether they would be funded or not. They didn’t necessarily get the money that quickly, but they knew they would. And then we moved to being four weeks and then we moved to six weeks over the course of sort of the 18 months that we were primarily running this. And there were other parts like bringing communities together to do research.

One way to describe that was if you want to understand the transmission of COVID, what you need in the room at the same time are a set of people who don’t usually work together. So you need some mathematical modelers, you need some people who do airflow in buildings and understand airflow, and then you need people who understand contact and social movements say across a bus or within an office. And then you need some people who are actually specialist virologist who can actually culture live virus to check whether it’s live or dead virus that’s being processed around. Trying to work out how you get those people together to very quickly start giving you answers and then change how people put their results out because that was the final part of the work was I can’t wait until you’re ready to write an academic paper about this. You need to start feeding this information back in to all of these places and making it publicly available much earlier than you would normally feel comfortable with, with all the caveats that go with that.

But it was literally, we need to say we need this information now and it’s no point in waiting for the perfect answer in four months. It’s the iterative process of improving that and it was kind of a crazy time I think trying to keep all of that moving, trying to keep all that information flowing, and trying to do these things. There are things that are easy to talk about like talking about helping get funding and early stage funding for some of the vaccines. A really fun one was about very early on made proposals that you can actually find out quite quickly if an area has a large amount of COVID by testing the sewage. And so working out how we could work with the various local authorities and you could find out where COVID was prevalent. Another one was collecting the stories of nurses who were working through the pandemic, because those are stories that would be very different if you asked them 12 months from now.

But it’s important to capture not just the data in terms of kind of physically what was happening in terms of virus, but also to understand the social effects of what we were doing. So we set off several studies in education during the pandemic whose results will come out many, many years from now, but you have to start collecting that data at that point. I’m afraid I could talk about this for a very long time because there were so many projects and so much that happened, but maybe that’s probably enough to give you a flavor.

Taren: It’s fascinating. If you had to pick one of those key learnings that came out of that, what would it be? Because I’m hearing collaboration across the board, an unprecedented collaboration.

Charlotte: I think there were several things, I think collaboration and openness in science. It proved just how powerful it was to be open about your science. You needed to tell people what you knew and so everybody did. Nobody thought, “Well I’ll just keep this sequence secret so I can have a bigger nature paper.” Everyone was like, “I need to tell everyone because more people need to know to be able to do this.” And we wrote into the kind of grants that we gave out here is kind of absolutely explicit. You are going to make your results completely publicly available because it was so important within this. The other thing I would say was this concept of sort of the iterative concept in terms of being able to build, so not the research so much but the way we were assessing the research and the funding principles we were using because of course the thing that we delivered over the first kind of four days, that wasn’t perfect.

Taren: Sure.

Charlotte: But you don’t wait four months to try and get it closer to perfect. You put that out and you tell everyone you’re going to change it and you learn from that and you just keep changing it and developing it. And I think that’s fed a lot into how I’m thinking about building the kind of biologic stuff here in Exscientia. This concept of, yes, you want it to be as good as possible, but if you don’t begin you don’t know what that is. And you also have to have a team of people who are happy with the knowledge that I will say in four months. Actually we’re going to change all of that because we can see it isn’t right and iterate onto a better version of the process and keep going in that way.

Taren: It’s certainly a different approach and I think it’s one that’ll prove to be far more efficient at the end. As you said, it’s iterative and it feels so much more transparent and it feels so much more future focused and brave, to be quite honest, because not everybody wants to work that way, so you really are creating something revolutionary. Kudos to you.

Charlotte: I’ll take that. You did ask at the end of the question what the recognition meant to me. I think I should answer that. It truly felt like the first time I had done something that my mom and dad were proud of. And what I mean by that is, and this this might be true for lots of people who do the kind of career I do, whose parents don’t work in the similar field and never have and there’s no reason why they should do, they don’t really understand what it means to me when I had papers published as an academic. But when they see my name on the list of the Queen’s honors list, they knew what that meant, I must have done something good and that was a really nice feeling and it felt like I’d given them a present. So, to be honest, for me the best part of that was it felt like I’d been able to give a present to my parents and it felt like they understood what I did a little bit at last.

Taren: That’s just lovely. They must be so proud. That’s amazing. Thank you for sharing that. You gave me a little chills there. That’s lovely. So talk about that since your parents are not involved in your field of study, what drew you to the life sciences originally?

Charlotte: I suppose honestly I’d say I wasn’t drawn to the life sciences originally, I was drawn to science, and perhaps even more if I think about it almost the physical sciences. This hasn’t changed in my entire life. I really love to learn. I love to learn as a small child, I love to learn now, and I really like to understand things and answer questions and I get really interested in problems. The minute somebody says ‘that’s not solved’ I become more and more interested in it. And I think the drive really came from it being a really interesting area to learning, so science I always just found it fascinating.

And I think the slow draw into the life sciences was actually because most of the really kind of, in my head, insoluble problems were sitting there and they were problems you could phrase in normal English. So there are plenty of insoluble problems in physics and other areas, but most of them they’re now so kind of esoteric and deep within the subject to describe even what the problem is. You come out of normal English and go into all sorts of things. Whereas if I sit here now, I can say, “Wouldn’t it be great if we could make drugs five times faster than we currently can?” and everybody sort of understands what I mean by that or “Wouldn’t it be fantastic if it was possible, if it cost half as much to develop a drug so it cost half as much for you to have one.” And being able to do that, that kind of link to the real world, I think has always been something I found really important.

And probably the last thing I should say, and this is obvious for my choice of being an academic as well, I really enjoy being able to tell other people about all of this as well. I find it fascinating to talk to people about kind of the problems in science and how they work and what happens with them.

Taren: That’s wonderful. Let’s talk about academics because you are still maintaining a role at Oxford and your career journey has spanned both public and private institutions. How has that all come together to inform your current role and why is it you noted that you still like to be able to explain the science to people. Is that why it’s important to you to maintain that connection to the university?

Charlotte: I think if I start the end there, that thing about the connection to the university, for me I absolutely love being part of the kind of academic research group. I suppose it sounds like a grandiose phrase but it’s kind of being able to train the next generation of people who work on my subject. And a lot of that is because I really like my job and I really, really want lots of other people to have that opportunity. So I feel it’s sort of incumbent on me to teach and train and show how much fun and how exciting it is and help them understand how they can come through to be able to do this. And I think all of that pulls onto why I decided that it’d be interesting to work within a private company as well because there are many things that you can’t actually answer questions within academia. There are differences in the kind of availability of various things or what’s important to do.

I think that all kind of translates into here. It’s also about how do I demonstrate this big change in the way we think about the process of designing a biologic as opposed to discovering one and how do I translate that idea into a reality that I can demonstrate and show. So, yeah, I think for me it’s sort of very similar, the two things, but the types of impact that I can have are very different. So in academia there’s this real people impact that I can see and feel and I really enjoy kind of…also just to be honest, doctoral students are a very challenging group of people to spend time with. They have lots of really great ideas and they have no preconceptions about what’s good or bad. So they’re a really great place to keep your brain very sharp and make sure you are really thinking about lots of really new and different things.

In Exscientia you are thinking also there’s a greater chance to be able to turn all of this into a reality, something that would actually kind of be a physical reality of being able to change the way that we do drug discovery which is something that academia, I can point the way but I cannot actually make that happen. So I think for me they’re all kind of interconnected by that desire to do those things.

Taren: And it’s interesting because it’s having a foot in both worlds, right? So in both as you said are complementary and additive to one another so that’s fantastic. You talked about giving back in terms of leading that next generation as scientists and you also are very much a dedicated mentor and that same kind of tone. So what is some of the best advice that you give to those who are looking to be a mentee or to be a mentor?

Charlotte: I find this difficult because I think every human is different so the general advice is…always feels a bit kind of, I don’t know the right word, a bit banal almost because the real important advice is about the fact that everyone is different and everyone knows nothing about something and something about something. So the kind of connections in terms of being a mentee and a mentor is if you’re a mentee, you’re looking for someone who you can see is in the kind of position or field that you would hope to be in one day, or you can tell has the kind of the experiences that can help you get there and that will change through your career on what is appropriate and what is helpful there. And some people never have a mentor and I think that’s fine as well.

On the other side, I think it’s that whole thing about if you agree to be someone’s mentor, you are agreeing to put time and effort into understanding what they want to achieve and putting time into talking to them about how they do that and where that comes from. For me it’s like all part of that sort of continuum of saying like I said with the students, I think it’s really important to be able to kind of see the next generation come through and how you help them become a better version. I sort of jokingly say to my PhD students it’s not totally a joke. By kind of training and being involved with them, I am far more likely to “win a Nobel prize” than I ever would myself. And by what I mean by that is if hundreds of students pass through my lab, the chances that one of them does something that really changes the world is much higher than the chance that I do personally. So I always think it’s a really obvious thing to try and train and help and support as many people as possible because the chances that real change will happen is made greater by doing that in my head.

Taren: That’s lovely. You’ve obviously been very, very successful in your career. Can you pinpoint to what are some of those key things that have made you so successful?

Charlotte: Thank you for saying I’ve been very successful. I guess in some ways I don’t know because it’s myself so you are always the hardest person to work out what you’re doing right and wrong as you go through.

Taren: Fair enough.

Charlotte: One of the things that I always think is quite a good piece of advice is when people are asking you to do things, it’s not that you always say yes or they always say no answer, but it is always start from the position of why wouldn’t I say yes because people tend to start from a position of, “Well I’m going to say no unless I should say yes” but start from the “Why wouldn’t I say yes? What is wrong with saying yes to this question to saying “Okay, I will try and do that.”

Another thing for me and I guess it’s been totally driven by my desire to learn is I have been very fortunate to kind of be able to be an academic at Oxford alongside as we’ve spoken a little bit about working with UKRI and now working with Exscientia and I think this broadening your horizons thing is…I mean I find it fascinating to learn how the different organizations work as well as the different things you can achieve within them. So not being afraid of stepping for me what turns out to be quite a long way outside my comfort zone at times to try and do things I had no idea if I could do them or not, but by just stepping there I can learn and if I learn then I’ll get better even if at that actual thing it turns out it’s probably not the right thing for me. So I think always that kind of thing about doing that is a great thing to try and do.

Taren: It’s exciting. I’m so fascinated by your duality of between the academic and then the science and then going into Exscientia and how you are really knitting all these different pieces together to, as we noted earlier, change how biologics are discovered and developed. Where do you see the science going next? Do you have a vision?

Charlotte: It’s quite a big question. I think we are at a point in time and I guess it’s in my head in terms of what would you train new employees or new students to be able to do now because you are preparing them for a career which is 10 or 20 years, these are skills that they should need in the future. And there are obvious clear things around this which is people need to have a much better understanding of the potential of AI itself but computation more generally, but AI in terms of what it can deliver and that is what I think will happen is that it will become a more commoditized property in the sense that you can get an AI algorithm. So it’s going to be all about that deep understanding of how you can use this in ways that actually add power to your decision making.

And another part of this I think will be around the explainability of these algorithms, truly having a statistical underpinning so we can explain the answers that they give and drive it back into a deeper fundamental understanding of if you like the physics and chemistry which are driving, if you like, the binding process between an antibody and antigen. In a sort of general sense, I think it really agrees with the kind of direction of Exscientia which is this idea of how do you encode and automate so much of what we currently do, can you do this better if you actually make use of all of these things. Another part of this I think is the rise of robotics within all of this. So we’re seeing more and more this kind of automation of labs that changes the types of data that you will get or the speed and the volume, but it also might change kind of what experiments you would think about doing because if you’ve automated part of the process, you now have people free to do more complex other experiments which I think will actually change the science of what we have going forwards.

Taren: Thank you for that. I love the part about the automating part of it and that’s a big people challenge too because when we start talking about automation and robotics, people are naturally fearful that their jobs might be eliminated or their skills might not be needed, but as you explained, it evolves them to that next level.

Charlotte: Yeah. I think it is a challenge because like all changes as you make them, people want to understand how they fit within that change. But to me what this is about is isn’t it exciting that you can use your skills to do something that you know something more than you were doing before. Because it’s not the case that we have solved these problems, we’re on the beginning of the journey of how you really change the way you do drug discovery. And so I think it goes back to my answer about that iteration thing. It’s continuously evolving and there will be more and more exciting pieces that we will need to do in order to make this work.

Taren: Charlotte, fascinating. Obviously throughout your career you must have had many moments that are like ahas or wows. So I’m going to challenge you to select one, maybe two, of those moments that either changed the trajectory of your career or have left a lasting impression on you.

Charlotte: I’m going to beg your forgiveness and have two. When I was kind of told about this, the first one that occurred to me and so that’s why I’m going to say it straight away, was I think the thing that actually changed the trajectory of my career is something that happened very early on and that’s when I got my undergraduate place at Oxford University. Because there are many parts around that I don’t think until then I had really understood that I would be able to do anything like this. And it was at Oxford I was exposed to all the things that were kind of potentially possible within science. And it was there that I really sort of thought actually “I want to do this, I want to be able to do all of these things.” And to be honest, it was also the place that gave me the confidence to do all of that. It’s like a little validation. It was that piece that says, “Yeah, you’re okay. You can be one of those people. You can think about this. You can build on this.” And so I think for me that was probably the thing that actually set me off on my whole track and if that hadn’t happened, I suspect I wouldn’t have done this.

The second one is something we’ve already talked about. When I went into UKRI to go and work with them, I went in as the deputy executive chair of the engineering and physical sciences council to kind of help them work with their strategic documents and areas they would focus on in terms of research and other pieces like that, talking to universities, designing and that sounded to me like a really interesting thing to learn to do and it was interesting for the first few months. But getting involved in the COVID response and I played a very, very tiny part in something that was massive and I got to meet some truly incredible people in that – both scientists who were doing the work, people who were attempting to manage all of this, people who were trying to corral all of these things. I think that, for me, also set me much more firmly on the track of if I can use my skills to do something which is if you like tangible and real, so this concept of can I build the real pipeline and not just write academic papers about it, that reminded me just how important that was and how hard it is to do something like that and so for me I think that was the other one. Sorry that there are two, but I feel they both really are those kinds of moments.

Taren: No, both are such wonderful stories and are so illustrative of your journey and I love that it was that spark at Oxford that gave you that sense of confidence that you could go on and really do remarkable things. And in the part that you played, and while I’m sure you’re being very humble in terms of the little piece, I’m sure it was incredibly important and really impacted hundreds of thousands of people. So two wonderful stories. Thank you for sharing them with us and thank you for sharing your journey with us as well as the exciting science that you’re pursuing at Exscientia. And I look forward to seeing the first biologic that comes out from your efforts, and I will be watching. So thank you so much for being part of our WoW podcast program.

Charlotte: Thank you very much for having me and it was great fun to have a chat.

Thanks for listening to this episode of WoW, the Woman of the Week podcast. For more WoW episodes, visit pharmavoice.com.