Meet Thierry Dorval: A 3-minute interview ⏳
Greetings!
This week, I spoke to Servier's Head of Data Sciences and Management Thierry Dorval about data management in drug discovery.
I enjoyed his take on the growing role of AI—with all its exciting possibilities and the risks.
Hope you enjoy reading,
—Luci, Senior Research Scientist, Synthace
👋 Meet Thierry Dorval, Head of Data Sciences and Management at Servier
Get to know Thierry better by reading the full conversation on Synthace's blog
Why did you get into drug discovery?
I didn’t plan to get into drug discovery. Originally, I was a physicist. I did a PhD in applied mathematics.
I only started learning about drug discovery and biology when I worked as a data scientist at the Institut Pasteur in South Korea. I began enjoying biology’s complexity, and I was humbled by it. I became interested in integrating biological understanding into thinking about what will be the next drug. At some point, my interest in drug discovery turned into an obsession.
What does a typical day look like for you?
As a leader and manager, I spend my time either organizing or being in meetings. It’s exciting, because I can go from guiding a PhD student on systems biology, to talking about company strategy with the research leadership team, to advising on how we can integrate new data.
Every day is an intellectual challenge.
In your view, what are the most pressing challenges in drug discovery?
We’ll have to think more about how we handle AI generated data in drug discovery, as over time there’s going to be a mix of synthetic data and experimental data—and there’s nothing that will help us tell the difference. We need to take this seriously, as in a few years, we could wake up and our data will be completely contaminated. We’ll face a crisis in trust.
What skills should all drug discovery researchers be developing?
1. Maintain a “scientific mindset.” You should only trust what you can test and validate.
2. Be resilient. There’s a chance that during your career, you may never be part of a success story. You need to come to terms with that.
3. Be open-minded. Because you don’t know what you don’t know.
Tell me about a recent project that excited you.
We’re currently building a knowledge graph for early-stage discovery, to bring together different types of data, either experimental or predicted, with different levels of trust.
This approach is super exciting, because the way you deal with the knowledge graph has nothing to do with querying a database: Instead you’re going from one connection to another. You’re deep-diving into the data and extracting knowledge in a very human way. And it’s very synergetic with our approach to AI. We’re using the combination in our internal project, and I think it will be the next step in data analysis.
What's a key thing that you've learned from success (or failure)?
Definitively, what makes things successful is collective intelligence. When you have access to chemistry, biology, and the clinical aspect, your project can become super efficient. If you're lucky enough to have this, make sure you tap into those resources.
What's coming in drug discovery that you're excited about?
I believe that at some point, de novo generation will explode and change the game. Whether it's de novo generation of small molecules, antibodies, or antisense oligonucleotides, we aren't starting from a lead. We're starting from de novo. If we combine this approach with experimental validation, I'm super excited about how things will evolve.
Any top tips for helping scientists adopt new technologies?
Keep an open mind, and avoid becoming that grumpy person who criticizes. Not knowing everything is what makes our jobs as scientists exciting. Take a hard look at yourself in the mirror, and ask yourself, “Am I as open as I was when I was a student?”
Who's helped you most in your career?
Jean Philippe Stephan. He's been a mentor to me, and he was the person who recruited me here at Servier. I wasn't a perfect fit for the position, but at the time I think that he saw something in me, and felt that I could do the job. Since then, it's been an exciting journey.
What are some recent books, papers, or articles you'd recommend?
Paper I’d recommend: Takahashi, K., & Yamanaka, S. (2006). Induction of pluripotent stem cells from mouse embryonic and adult fibroblast cultures by defined factors. Cell, 126(4), 663–676. DOI: 10.1016/j.cell.2006.07.024
Book I’d recommend (French): La Sculpture du Vivant: Le suicide cellulaire ou la mort créatrice by Jean-Claude Ameisen
🧊 The 2 types of lab work in drug discovery
According to Markus Gershater, there's the high-throughput, high-volume work—the easily automated and standardized kind. Then, there's the highly complex and variable lab work you have to do before you can get to that point.
💊 More candidates ≠ more cures
Drug discovery and automation specialist Daniel Yip rightly points that AI flooding our pipelines with more candidates can only take us so far—especially if the real bottleneck is experimental validation.
🫵 Your lab robot needs you...
To tell it how to pipette your liquids. Field Application Scientist II Nathan Hardingham encourages everyone to pass on their instinctual knowledge gained through hand-held pipetting by setting up liquid classes.
🔗 Content we're loving
The structure of human sweetness
Why we loved it:
Scientists have cracked the 3D structure of the human sweet taste receptor—the dynamic GPCR duo TAS1R2 and TAS1R3—using cryo-EM. TAS1R2 does the heavy lifting by binding sweet molecules like sucralose and aspartame, while TAS1R3 stabilizes the protein complex and helps relay the signal.
The team uncovered a shared binding pocket and coordinated structural shifts between the subunits, offering new insights into how we detect sweetness at the molecular level. It’s a major step forward for understanding flavor perception—and a promising path toward engineering better, smarter sweeteners that can satisfy without the sugar overload.
Why we loved it:
A research team supported by the National Institutes of Health (NIH) has used personalized CRISPR-based therapy to successfully treat an infant with carbamoyl phosphate synthetase 1 (CPS1) deficiency, a rare and life-threatening urea cycle disorder.
The treatment, developed and delivered within 6 months of diagnosis, corrected a specific mutation in the CPS1 gene in liver cells—restoring partial enzyme function and reducing toxic ammonia build-up in the body. As the first use of a fully personalized, somatic CRISPR therapeutic in a human patient, it’s a major leap for precision medicine. It goes to show that gene editing platforms can be rapidly adapted to target other rare genetic diseases.
About Synthace
Get faster, smarter insights from your R&D experiments. Designed by and for biologists, Synthace lets you design powerful experiments, run them in your lab, then automatically build structured data. No code necessary. Learn more.
Get the latest from our lab team
Scientists Luci and Emily help you absorb wisdom from experienced folk in drug discovery. Plus, they give you the latest on automation, multifactorial methods, and effecting change in life science R&D.