Meet Casper Wilstrup

CEO at Abzu

We met Casper for a conversation about scientific frontiers, the programming language of DNA, and why drug discovery is so expensive.
Abzu® is the startup behind the discovery engine the QLattice® and the discovery system Reason® that accelerate exploration, enabling scientists to bring drugs to market faster.
Play Video about Casper Wilstrup interviewed on Abzu Discovery Loop
Abzu challenges the fundamental assumptions of black-box AI with its pioneering research engine, the QLattice, to bring drugs to market faster.

“You never know where startup life will take you.”

Casper has worked for and founded startups ever since his youth, but never expected to be in the life sciences or pharma.

“I’m on the physics and mathematics side of things, typically, but life science is a new one for me. But it is a place where analyzing data makes a lot of sense.”

– Casper Wilstrup, CEO at Abzu

“Is life written in a programming language called DNA?”

“It can be that the consequences are hard to detect, or maybe they’re not particularly big in a given context or in a given time frame, but there are no components to a physical system that have no consequences. And therefore, of course, ‘junk DNA’ has consequences – although not as linear and directly as perhaps coding parts of DNA.”

– Casper Wilstrup, CEO at Abzu

“So what a treasure trove of potential druggable targets lie in the 80% considered ‘junk DNA'.”

As an 'outsider', Casper was surprised by the drug discovery process.

Casper Wilstrup tells the story of Abzu on a desk
“There were two things that surprised me, and therefore also two things that I think it’s important to tackle. And I also think it’s possible to tackle them.
You want to make better decisions about what to try, and you want to fail faster with the things that are wrong.”

– Casper Wilstrup, CEO at Abzu

Why is drug discovery so expensive?

“It’s decisions about what to try, what to do, driven by actual scientific knowledge. And finding a way to test the things that you believe (but which you’re uncertain about) at as early-as-possible a state where failing is cheapest.”

– Casper Wilstrup, CEO at Abzu

Casper at TechBBQ

Drug discovery should be a loop, not a funnel.

The Discovery Loop series is for those who share the vision that through a smarter organization and better utilization of data, we can make drug discovery better, faster.

Stay in the loop

Sign up to receive updates. You can opt out at any time.

We’re cookieless, and our privacy policy is actually easy to read.


Martin: Casper, why don’t you introduce yourself?

Casper: I certainly can. I’m Casper. I am an entrepreneurial spirit who has worked ever since my youth on a sequence of startup companies, the latest one being Abzu, but also by far the one dearest to me being Abzu. I have worked mainly with high performance computing and data analysis for the last 25 years in different contexts. That kind of thing has now acquired the name artificial intelligence. It’s still data analysis. And here at Abzu, we are taking a leap into a completely new approach to data analysis that I’ve been thinking about for more than 25 years.

Martin: And you found yourself in the life sciences sector maybe for the first time?

Casper: That’s the interesting thing about life, and I guess startup life, is that you never know where it takes you. I had no idea I’d end up in the life sciences or in pharma, actually. I’m on the physics and mathematics side of things, typically, but I’ve worked in many different industries, from shipping to marketing to all sorts of industries in my career, but actually never life science. It’s a new one for me, but it is a place where analyzing data makes a lot of sense.

Martin: Yeah, for sure. I’ve been in life sciences for all my career, in fact, and so I’m having the slight switch in terms of working with such interdisciplinary people. Well, yeah, I’ve dabbled in that, but we’re part of a really diverse group of people and great diversity and skill sets. I was kind of pondering on how I arrived here a little bit when we were planning this discussion.

And I was thinking, when I was an undergraduate, there was no such thing as the human genome. It hadn’t been sequenced. And that happened about midway through my PhD, which was in life sciences. And I felt something seismic was about to change in the way that we approached drug discovery, new medicines, and understanding diseases. And that’s 20 years ago.

And now I think we’ll probably touch on the phrase omics, polyomics, multiomics. So that was the genomic era era, and now we’re in the multiomic era, which, as you’ve alluded to, involves lots of data. And I think pharma, for one, biotech, the whole life sciences sector, are realizing how important it is to embrace that. To embrace the wealth and diversity of data. But there are problems there and maybe we’ll talk about some of the problems around data in pharma and life sciences.

Do you have any initial thoughts or comments on what the big issues are around?

Casper: Let me first say that I’m with you. I was also apprehensively waiting for the human genome to be sequenced back in the zeros. Even though I wasn’t in life science, I’m a curious person, so I was following it carefully, just like I’m following everything that the James Webb telescope is doing these days.

And I had the same feeling – not as informed, I’m sure, as where you came from – but back then I had the feeling that life is written in a programming language called DNA. So now that we have reverse-engineered that programming language and laid it bare in front of our eyes, we would understand.

I wasn’t naive enough to think that we’d understand all about life and how it evolved, but I had thought that we would understand more than what has come out of sequencing the entire genome. I think from an information theoretical point of view, I was surprised, but I shouldn’t have been thinking about information theory as it really is and how complex systems interact.

Of course, you couldn’t explain biology from the DNA, but that just means that we didn’t get the benefits that we immediately thought we’d get. But we got a lot of very interesting things to study and try to understand that is leading its way into new science. And for me, being part of that new science, it is a scientific frontier in a certain sense to understand, even understand what kind of data we should collect and what data means what and what data in a certain sense even is when you’re talking about biology, it’s extremely fascinating. And it’s right where we are.

Martin: Yes, 20 odd years ago that the human genome was published. And we were just having a conversation earlier today about the fact that 80% of that was kind of thought of as junk, just carrier DNA, not doing anything. And it’s 20 years on we’re really starting to get a better understanding about why there’s 20%. That’s really important. But 80% is highly involved in intricate regulation of that 20% and it opens up the whole drug discovery sphere, really, to think beyond the 20% that makes a protein that you can target as a drug, which I think is the sort of traditional drug discovery route that people have taken making small molecules or antibodies to these proteins. And 80% are still unexplored and are just starting to be explored. I think that’s kind of mind blowing.

Casper: There’s a lot we could say about that, right? So first of all. Yes. Even the term junk DNA. From a certain perspective, I can see why you would come up with the theory that the stuff that gets converted into protein is the stuff that matters and the rest is just there for evolutionary leftovers but on the other hand. In any system that you study, maybe physicists speaking, in any physical system that you’re studying there are no inconsequential components. Right? It doesn’t exist. It can be that the consequences are hard to detect or maybe they’re not particularly big in a given context or in a given time frame but there are no components to a physical system that have no consequences. And therefore, of course, the junk DNA has consequences although not as linear and directly as perhaps a coding part of DNA.

But I think that’s also what you’re pointing to here that most of drug development today, as I see it at least – and of course you’re the bigger expert – but as I see it, most of drug development, because we’re using antibodies and small molecules, really are limited or constrained to targeting the stuff that gets turned into proteins and that gives it cuts off a lot of causal threats or avenues of action in the organism that we just cannot attack with those kind of drugs because it’s after it gets transcribed to proteins. So what treasure trove of potential targets really, druggable targets lies in the 80% and that of course brings us to RNA therapeutics that we both care about deeply I know.

Martin: Maybe it’s a good time to go into the RNA therapeutics world. I did research before moving into business development and before Abzu, and I heard from pharma companies time and time again: We need better disease understanding, we need academia to find us new targets because there’s lots of money in drug development but we’re all chasing the same targets and that’s a phenomenal waste of money.

It’s not serving vast numbers of patients for which there are diseases with absolutely no medicines on the horizon and things just need to be changed. The whole way about tackling drug discovery is a bit outdated and I think the RNA therapeutics revolution that is starting is one way of opening up those opportunities. And I would say one of the beautiful things about RNA therapeutics is that it marries itself so well with omics data sequences. These drugs are precision medicines; they’re designed down to single nucleotide precision and combining understanding genetic evidence, proteomics, all these different omics, being able to put them together and start understanding what’s going on will lead to thousands of new medicines for thousands of diseases that as yet have got no treatments and I think that’s really exciting.

Casper: There is still the topic about… So one thing that when I first learned about RNA therapeutics, specifically antisense oligos, well, here’s the silver bullet: You just define the reverse transcript of whatever mRNA or pre mRNA you want to knock down and voila. And then I realized, right, but that requires it to be stable, to be deliverable into the nucleus, in this case, to have no adverse side effects of any kind, to be perfectly safe. And of course, those are the other details that hide the devil.

And in that sense, I think RNA therapeutics is in its nascency. There’s a lot of things we don’t really understand about the way it also impacts the body other than its direct impact for which it’s designed to. So it can mean that you get this feeling of there’s a long way to go. But for me, it’s more like it’s a treasure trove of undiscovered discoveries. And that fascinates me in the field.

There’s also another thing about why I’m really fascinated with RNA therapeutics as a modality, which is that it opens the opportunity for the scientific method being applied for target discovery. Because what’s so hard about target discovery is that it really is just a hypothesis, right? You dig and you study and you read and you measure and you analyze and you compute statistical probabilities. And then you end up with a belief that this target plays a certain role in the development of a disease.

But in honesty, it really is a fairly unsubstantiated hypothesis at this stage in almost all situations. So for a scientist, I think the next natural question would be: What is the easiest possible way I can falsify this hypothesis? The answer is right there. Right? RNA therapeutics can readily test your hypothesis, at least if you’re able to express it in a way that can be tested in a cell line. But no matter what, it is the easiest way you can possibly think about to validate or falsify a hypothesis about a target playing a certain role in a biological mechanism. And that makes it a wonderful tool and a very underutilized tool, I think, in the industry.

Martin: So we have been talking about the traditional way that drugs have been developed in pharma and biotech and that being a very linear process. So you have an unsubstantiated target, you try your best to validate it, and you’ll do that in vitro, in cells, in a dish. And the ones that still look good then go into small animals and then to bigger animals before they ever get to clinic to test the first man studies, as they might be called, phase one studies for toxicity.

That is a very long and a very expensive journey. I think the attrition rate is something like 80-90% and the cost to go from ‘idea’ to ‘testing in clinic’ to ‘drug on the market’ can vary between, traditionally between, $1 and $10 billion. It’s insane. And so if 90% of your drugs are failing, something needs to be done. And we have the idea, I think, collectively that with the information that we can gather from these RNA therapeutics and with data that comes out, we can start to integrate all the information that we’re getting and feed it back continuously, which we’re terming a drug discovery loop, an omics drug discovery loop.

I’ve not turned that into a question. That’s just a ramble.

Casper: No, but it’s a good conversation. I’m very interested in this. I can jump in and say again: As an outsider coming to the field less than two years ago, I knew next to nothing about disease mechanisms and how the drug development process actually worked. So there were two things that surprised me and therefore also two things that I think it’s important to tackle and I also think it’s possible to tackle them. And I think the best way to phrase it is, at least for me, a good way to phrase it is in the context of another field that I’ve worked.

I’ve worked a lot in software development where when you develop software, you want to make better decisions about what to try and you want to fail faster with the things that are wrong. That’s really the two pieces. You want to get more software out the door, more workable, usable, nice software out the door. You try to become better at making good decisions about what to build, but acknowledging that it is just speculative. You also try to find out: How can I fail fast if I’m not building the right thing?

Now, translate to drugs. It’s decisions about what to try, what to do. Driven by actual scientific knowledge is the ‘make better decisions’ part. And ‘finding a way to test the things that you believe, but which you’re uncertain about, as early as possible a state where failing is cheapest’ is the other part.

I think maybe when you say the “odd loop,” that’s what you mean. You mean being able to make informed decisions, also make informed decisions about how to test what you still don’t know, test what you still don’t know, and then bring that back, particularly when it fails. Then you don’t progress in the loop. You go back. But now you go back with knowledge that makes you take more informed decisions in the next try.

So ultimately the loop just spins faster and faster because the knowledge builds up. But in the knowledge build-up is the reason for – so there’s something about how you organize yourself to achieve this faster loop, which is cross functional teams and empowered teams and those kind of things – but that’s one part. The other part, the thing that spins the wheel faster and faster, is actually the accumulation of knowledge.

Martin: Well, that’s an interesting point. So individual entities, organizations, pharma, biotech, could be doing this themselves and keeping what knowledge they have to themselves. But there is a wealth of information globally, and if there was confederated data of some description, would that make everything work?

Casper: It’s one piece, but ultimately it’s not about data, it’s about understanding.

Returning to the spin faster and faster metaphor, it’s the understanding that makes the wheel spin faster and faster. It’s not the data. But the data can facilitate the understanding. It doesn’t come naturally.

Actually, that’s where data analysis comes in, right? Data analysis and scientific analysis of data is about converting data into knowledge. So if it’s only the accumulation – rapid accumulation of knowledge – part of the challenge, yes, data is part of it. But analyzing that data in appropriate ways and interpreting what comes out of that analysis and committing it to your understanding, to your common domain of understanding, is equally important.

I’m a big, huge fan of the idea of consolidation of data in various ways in a federated setting where data stays where it is, but you can learn from it in a distributed fashion or in a centralized setting where there’s some kind of shared access. Either way, what you get is the ability to extract knowledge out of the data. But even if we fail at that, all hope is not out because if you can extract knowledge from the data, then that knowledge can be committed to the shared pool without the data ever being shared in any meaningful sense.

I think in a way, that’s how the history of science has actually progressed. Sometimes we repeat each other’s experiments, but most of the time we build on top of each other’s conclusions and understanding. So there’s a differential learning in that sense happening in science as such, which I think is also the hope here. All right.

Share this loop

Stay in the loop

Sign up to receive updates. You can opt out at any time.

We’re cookieless, and our privacy policy is actually easy to read.

Contact Abzu

It’s easy to get in touch with an Abzoid.

Contact form

Fill out the form below, and an Abzoid will be in touch in 24 hours.

Email us

Email a scientist at or

Call us

Reach us during standard CET business hours at +45 31 23 47 64.