Founder Spotlight #27: Nicolas Tilmans @ Anagenex
BIOS: Nucleus of Life Science Innovation 🚀
- BIOS Talent: Find Jobs @ Breakout TechBio Startups — Search Jobs 🚀
- Post Jobs: Add Your Startup to BIOS Talent — Post Now 🎉
- Students: Join Alix Ventures Fellowship — Join Now 🧬
- BIOS Contributor: Share Your Thought Leadership — Join Now🔬
CONTENT & COMMUNITY
- BIOS Daily: Join 25K+ Subscribers Following TechBio — Sign Up 🔥
- BIOS Insider: Premium TechBio Thought Leadership — Sign Up ✨
- BIOS Commons: World’s Largest #TechBio Community — Join Now 🎉
- BIOS Angels: 1st TechBio Angel Investing Syndicate — Join Now 🌟
- Alix Limited: Invest in Breakout TechBio Startups — Learn More 🧠
Alix Ventures: Supporting Early Stage Life Science Startups Engineering Biology to Drive Radical Advances in Human Health
Calling All Innovators — Click to Reach Out 🚀
Nicolas Tilmans is Founder & CEO @Anagenex, a company applying massively parallel biochemistry and machine learning to find new medicines. He has spent most of his career alternating between bench and computer science, studying biochemistry and computer science as an undergraduate before doing a Ph.D. in Biochemistry at Stanford. There, he helped to develop a flavor of DNA Encoded Libraries (DEL), a technology that increases the throughput of early-stage drug discovery 1,000 fold (this work would eventually be licensed by a company called DiCE Molecules). After graduating, he worked as a data scientist in industry, eventually becoming the VP of Engineering at a patient data machine learning company before founding Anagenex.
Anagenex is a seed-stage biotech combining very high throughput biochemistry with machine learning (ML) to discover new treatments for hard diseases. We believe that combining the strengths of both technologies, namely generating (DNA Encoded Libraries, aka DEL) and processing (ML) large volumes of data, has the potential to identify new chemical matter faster and more reliably than competing approaches. More specifically, we use novel selection strategies to probe beyond simple binding, and apply machine learning to move beyond the limitations of DEL compounds. By creating a totally integrated computational and wet lab platform where both disciplines feedback into each other, we find compounds that modulate challenging disease targets then efficiently optimize those molecules to become drugs that succeed in the clinic.
What prompted you to pursue a career in Healthcare?
The first time I became interested in biology was in the 5th grade. Time Magazine had just published a genome piece highlighting Craig Venter. It was around the time that the human genome project was launching and starting to get some press. It sounded super cool, so I decided that I wanted to be a geneticist.
Later, after taking some chemistry and biology classes in high school, I figured out that I actually wanted to be a biochemist. At the same time, the internet was becoming more available, so I joined a student-run team that built and maintained a rudimentary network at the school, and taught myself the basics of Linux, HTML, and C++ after class. I enjoyed it so much that I added a Computer Science major to my Biochemistry plans in college and since then I’ve always sat at the intersection of biochemistry and computers.
I wanted to do impactful work, and drug discovery is a field where one can legitimately say that success saves lives. We desperately need more medicines and we need to find them fast. I think that combining computers and lab experiments the right way can create a virtuous cycle where each can feed back into one another and make the drug discovery process as robust as we all need it to be.
What problem is your company solving? How did you become motivated to tackle this particular problem?
Everyone we know including ourselves will someday be sick, and too often medicine has no response. Worse, sometimes we have a good idea of what might be responsible for the illness, for example, what protein to target as the culprit, and yet we remain powerless.
No disease should be immune from interrogation and therapeutic modulation.
We have fallen short in the past because even when we know what biology to target, finding candidate medicines is very unreliable and takes forever. A new medicine costs $2B and 5–10 years of work to develop today. That’s too much and too slow.
Of course, clinical trials are a necessary and major cost for drug discovery, but 40% of the cost comes from the early stages, chiefly:
- Finding a good molecular starting point for a medicine (a “hit’)
- Optimizing that compound into a drug (“lead optimization”)
At Anagenex we are solving this problem by running massively parallel biochemical lab experiments and using the results to teach machine learning how to make better drugs. We iterate between lab and computation over and over, refining our understanding of what compounds could be good medicine for a particular disease target.
In essence, we make a large population of molecules, see which ones survive a fitness test (does it interfere with a disease-relevant protein), create a new population-based on the fittest molecules, and iterate the process. We are developing an artificial evolutionary process to find new drugs.
In doing so we widen the number of problems we can tackle, increase the probability of success and reduce the cost of the inevitable failures.
Now in technical language, what are the specifics of what your company does?
We solve the molecular recognition problem for small molecules, reliably and quickly.
At Anagenex we use a technology called DNA encoded libraries (DEL) to test billions of compounds in parallel for binding to a protein we think might cause disease. The results teach a machine learning model what compounds could be good drugs.
At this point, instead of just asking the machine learning model to tell us what are the 20 best compounds to make, we ask the model where it’s more or less confident in its predictions. Based on that information, we then make a new set of about a million molecules to refine our understanding of what compounds could be good medicine for a particular disease target.
By iteratively running this process we:
- Find more chemical starting points for a given target
- Rapidly optimize those compounds into candidate drugs
- Test them in disease model systems
- Fall back to one of many backup starting points if the compound doesn’t work
We focus on several types of challenges. First, we try to target the protein targets which are known to be important but historically have proven very challenging for drug development, such as DNA- and RNA-binding proteins. We also work with clinically validated targets where small molecule modalities could be more effective or convenient than the current drugs, for example, substituting for the antibodies.
Where are you now in building your technology?
We have now proven that our platform works. We are now moving to apply it to a series of more difficult problems including a few synthetic lethal cancer targets that require highly specific compounds and nucleic acid binding proteins
We have shown that we can:
- Build an incredibly clean DEL full of molecules that can eventually become drugs
- Use proprietary low-noise selection strategies to identify good molecules in those DELs
- Train a machine learning model from that data, and have it identify new molecules
- Build a new DEL of roughly 1 million compounds in under 3 weeks based on the ML model’s predictions
- Take data generated with that new DEL and update the model to get better predictions
We are now pointing this platform at difficult problems in drug discovery such as selectivity (the drug can only interfere with one protein in a family with many close relatives), challenging target classes (such as nucleic acid binding proteins), and allosteric binding pockets (binding to a protein outside its normal active site).
What is your company’s founding story?
After grad school, I had taken a break from biotech and went back to computer science for a while. I met up with Zavain Dar (GP @ Lux Capital) for dinner one night and we had a conversation on directed evolution and drug discovery. Founders don’t talk enough about how scary it is to start a company and to find the right people. I got lucky that I had the right kind of support early on. Zav was a real partner throughout the entire thought process and still today. Eventually, I kept thinking about this idea and concluded that it had to work. We went out and raised capital and the rest is history. We were lucky to have everything fall into place relatively quickly, though it hasn’t been without challenges. For example, COVID made us rethink how and where to build the company. We were figuring out our first hires when the pandemic hit, and all of those plans had to be reworked. We’re now a hybrid company with labs in Boston and a remote computational team.
What are some of the notable milestones your company has achieved thus far?
We went from an empty lab shell in November of 2020 to proving our system from end-to-end by November 2021. Our lab team leads, Joe Franklin and Svetlana Belyanskaya have been doing this for at least 15 years and redesigned the DEL framework from the ground up to be the best it has ever been. Our ML team led by Henri Palacci has designed new model architectures to take advantage of our unique data. Seeing the first molecules coming from our integrated ML-Lab system confirmed in experiments was a special moment.
On a personal note, I’m proud of watching the team grow together. There was one moment where we were all at a whiteboard, breaking down a paper. Joe had one idea, Polina (our Machine Learning Scientist) had another; by playing off each other’s strengths, we ended up coming up with a solution that looks at the problem completely differently. Traditionally the lab and computer elements have been siloed, so seeing us move as one team is incredibly gratifying.
What are some of the biggest hurdles ahead?
Continuing to have the computational and experimental parts of the company communicate smoothly takes effort. People come from very different backgrounds and different cadences. Wet lab researchers are very often focused on one experiment which is a physical process. Computer scientists can run dozens of virtual experiments in parallel and generate a huge number of hypotheses. It’s important for computer scientists to understand the physical constraints of experimental biology, and it’s important for lab scientists to trust the results from computation even if they’re counterintuitive. Having said that, we have designed the company to marry the two worlds together from the very start.
Pay It Forward
Throughout the journey, what have been some of your biggest takeaways thus far?
Luck matters a ton. You can make your own luck and figure out ways to increase your potential for getting lucky, but it helps to have things go your way and you’re not always in control of whether they do. Reading widely, talking to other founders, investors, academics and other members of the startup ecosystem all help create that luck.
What are some of the must-haves for an early-stage Healthcare/Life Sciences startup in your eyes?
Big pharma and biotech companies have way more money than you do, but they can’t move as fast or see a new opportunity as clearly. Play to those disadvantages, clarify what you are doing in your process that isn’t relatively easy for a bunch of smart, highly funded people in a big organization to do.
Platforms are valuable only if they can reliably generate drugs that will get to a patient. The drugs you find will be more valuable than the platform itself, so always have a clear understanding of how you will get to something that is valuable for patients.
This is a conservative industry, and for good reason. Think hard about where that conservatism is warranted and where it is outdated.
What are some of the traits that make a great founder? What type of risk profile/archetype does someone need to have to be a founder in your opinion?
Humility is an under-appreciated skill in founders. My interdisciplinary training gives me an edge in understanding how computation and lab experiments can reinforce each other when tightly integrated. In all other areas, there are people in the company who are smarter and better understand the details of our processes. Never let ego get in the way of the right decision.
I’d say that most people think startups are riskier than they actually are. If you try a startup for a year or two and it fails, chances are you’ll find the same (or likely a better) job than you would otherwise. If it succeeds, then you’re way better off anyway. If you strongly believe the idea has a chance, take the risk.
Not everyone knows everything. Often founders have to learn either the science or the business side better. What advice would you give for someone picking up a new skill set such as this?
The most important skill for a deeptech founder is to be able to communicate technical ideas extremely clearly. That is your secret weapon. If you can’t communicate, it doesn’t matter what you are doing or how strong you are technically. My first decks were torn apart for having too much science, even though I thought I cut a lot out. I still struggle with this constantly as I present to different audiences.
Can you demystify the process of what it was like to raise VC funding? What were the highlights & low lights? Any advice or words of wisdom for future founders?
Focus on the key story. VCs need to understand the key idea within the first minute or two of your pitch. You probably understand the details better than anyone, so get them hooked so you get a chance to tell them.
Our process was pretty smooth, but we still got told no a whole bunch of times. It’s hard not to take it personally, but you have to push through it.
What advice for managing and hiring a great team can you share?
On hiring: Find people who have something to prove. If that thing is well aligned with what the company needs, you’ll be unstoppable.
It’s cliche but hiring smart, competent people is critical. I’d go so far as to say it’s maybe the only thing that matters. Not quite (you do need a decent strategy), but almost.
On management: Management comes down to empathy and motivation. People don’t burn out because they work too hard, they burn out when their hard work doesn’t go anywhere. As a manager, you have to keep a pulse on how people are connected to the work and how they are feeling.
You’ve hired smart people, trust them with the freedom to execute. If you feel like you have to micromanage them, you probably don’t have the right person. If you micromanage a person who is highly competent, they’ll probably leave.
Finally, you can’t repeat the important things enough. People often won’t hear you the first time, you have to repeat it all the time, often in different ways so that everyone stays on the same page.
Any closing words of wisdom?
In the vast majority of careers, your life is about taking a test and succeeding to the next stage. That is not true as an entrepreneur. One of the hardest things as a founder is dealing with how many decisions have no clear answer. You will get advice, all important, some of it contradictory, but at the same time, all of it right. Your job is to make peace with that. Figuring out how to make decisions that have no clear answers and living with that is one of the hardest and most valuable things a founder can learn.
The good news is that, again, this isn’t a test where you get a grade and there’s no going back. Most decisions can be reversed or modified, so just pick a path and move. Speed is often more important than perfection.
📣 If you enjoyed this post please clap 👏 & comment 💬 to let us know
❤️ Thanks Sasha Eremina for your help in putting this together :)
Alix Ventures, by way of BIOS Community, is providing this content for general information purposes only. Reference to any specific product or entity does not constitute an endorsement or recommendation by Alix Ventures, BIOS Community, or its affiliates. The views expressed by guests are their own and their appearance on the program does not imply an endorsement of them or any entity they represent. Views and opinions expressed by Alix Ventures employees are those of the employees and do not necessarily reflect the view of Alix Ventures, BIOS Community, affiliates, and content sponsors.
Join BIOS Community 🎉
Become a member, continue the conversation, connect with like-minded Life Science innovators, access exclusive resources, & invite-only events…
Apply to Join — Membership Application